The Duty of Responsible Administration and the Problem of Police Accountability

Charles F. Sabel and William H. Simont

Many contemporary civil rights claims arisefrom institutionalactivity that, while troubling,is neither malicious nor egregiously reckless. When lawmakers find themselves unable to produce substantive rules for such activity, they often turn to regulating the actors' exercise of discretion. The consequence is an emerging duty of responsible administrationthat requires managers to actively assess the effects of their conduct on civil rights values and make reasonable efforts to mitigate harm to protectedgroups. This doctrinalevolution partially but imperfectly converges with an increasingemphasis in public administration on the need to reassess routines in the light of changing circumstances. We illustratethe doctrinal and administrative changes with a study of policing. We discuss court-supervised reforms in New York and Cincinnati as examples of contrasting trajectories that these developments can take. Both initiatives are better understood in terms of an implicit duty of responsible administrationthan as an expression of any particularsubstantive right. The Cincinnatiintervention, however, reaches more deeply into core administrativepractices and indeed mandates a particular crime control strategy: Problem-Oriented Policing (POP).As such, it typifies a more ambitious type of structural intervention that parallels comprehensive civil rights initiatives in other areas.

I. Introduction ...... 166 II. The Evolution of Civil Rights Doctrine ...... 169 A. The OrganizationalPremises of Classic Doctrine ...... 170 1. A ntidiscrim ination ...... 170 2. Search and Seizure ...... 172 3. Attribution Under Section 1983 ...... 173 4. Structural R elief ...... 173

t Maurice T. Moore Professor of Law and Arthur Levitt Professor of Law, respectively, Columbia University. We are grateful for advice and encouragement to Jeffrey Fagan, Bernard Harcourt, James Liebman, Tracey Meares, Gillian Metzger, Joanna Schwartz, David Sklansky, Samuel Walker, Robert Weisberg, Theodore Wojcik, and to many participants in Cincinnati police reform who spoke to us about their work, including Jeffrey Blackwell, John Eck, David Epstein, Alphonse Gerhardstein, Daniel Gerard, Mars Herold, Joe Lorenz, and Michael Neville. Yale Journal on Regulation Vol. 33, 2016

B. The Emerging Duty of Responsible Administration...... 176 1. A ntidiscrim ination ...... 178 2. Search and Seizure ...... 180 3. Attribution Under Section 1983 ...... 181 4. Structural R elief ...... 183 111. The Evolution of Policing ...... 185 A. From Reaction to Proaction ...... 185 B. Two Trajectoriesof Reform ...... 186 1. A ssertive Policing ...... 188 2. Problem-Oriented Policing ...... 193 IV . Civil Rights and Police Reform ...... 201 A . Com p liance ...... 202 B. The H olistic Approach ...... 206 1. Limits of Specialized Reform ...... 206 2. Problem-Oriented Policing as a Civil Rights Remedy ...... 209 V . C onclusion ...... 2 12

I. Introduction

Public officials have a duty of responsible administration that consists of three elements. It requires that officials articulate reflectively the policies and principles that govern their work. It demands that they monitor the activities of peers and subordinates to induce compliance with these policies and principles. And finally, the duty mandates frequent reassessment of these policies and principles in light of the officials' own experience and that of comparable institutions. The "duty of responsible administration" is our name for some converging trends in constitutional law, common law, and statutes. The term usefully connotes developments across a range of fields. It resonates with interpretations of the constitutional Due Process and Take Care Clauses that entail obligations of general proactive administration.' However, the most important recent authority for the duty arises from efforts to elaborate provisions of substantive civil rights law. Where courts or legislatures cannot mandate specific substantive directives, they often turn to regulating the ways officials give content to their discretion. Recurring procedural themes in the elaboration of various substantive doctrines suggest a set of implicit overarching norms. In a reversal of a process

1. Jerry L. Mashaw, The Management Side of Due Process: Some Theoretical and Litigation Notes on the Assurance of Accuracy, Fairness, and Timeliness in the Adjudication of Social Welfare Claims, 59 CORNELL L. REV. 772 (1974) (suggesting that "due process" entails general proactive administrative duties); Gillian E. Metzger, The Constitutional Duty to Supervise, 124 YALE L.J. 1836, 1875-86 (2015) (considering the Take Care Clause of Article 11 as a source of a duty to supervise subordinate officials). Police Accountability noted by Henry Maine, procedure has been secreted in the interstices of substance.2 At the same time doctrine is becoming more procedural, administrative processes are evolving. Agencies have been moving away from bureaucratic forms of administration. Bureaucracy, as understood in mid-twentieth century America, was a balance of stable, hierarchically promulgated rules and lightly supervised discretion. Yet, this kind of organization no longer seems appropriate for many contemporary problems. Addressing current problems requires both more flexibility than rules permit and more transparency than discretion typically affords. Efficacy depends on frontline initiative but also demands that such initiative be reflective and accountable. Thus, administration is drawn to post- bureaucratic forms of organization that emphasize provisional and easily revised plans, monitoring designed to induce learning as well as compliance, and systematic reassessment on the basis of experience within the agency and in comparable institutions. These developments in civil rights doctrine and in public administration originated independently of each other, but they now converge strongly. Second- generation doctrine turns substantially on notions of reasonableness, and reasonableness tends to be assessed in terms of accepted patterns of administrative practice. Moreover, where administration is bureaucratic, courts must either intervene by imposing rules, which would often inefficiently rigidify administration, or defer to unaccountable discretion, which would often leave lawless practice unremedied. Post-bureaucratic administration makes possible responses that do not carry the disadvantages of either of these courses of action. We illustrate these developments by a discussion of civil rights law, especially as it relates to policing. Scholars have noted the administrative turn in civil rights doctrine.3 The classic Warren and Burger Court era cases have proven inadequate to address many "second-generation" problems. First-generation problems typically involved intentionally harmful or egregiously irresponsible conduct. Classic doctrine often defined liability in terms of individualistic psychological notions such as "discriminatory intent" or "deliberate indifference" and prescribed remedies in the form of bureaucratic-type rules. By contrast, second-generation cases often arise from unreflective or normatively ambiguous conduct that, although troubling, does not fit the psychological premises of classic doctrine. Legislators, judges, and regulators often find that

2. HENRY SUMNER MAINE, DISSERTATIONS ON EARLY LAW AND CUSTOM 389 (1883) ("in the infancy of Courts of Justice,.. substantive law has at first the look of being gradually secreted in the interstices of procedure"). 3. See generally,Samuel R. Bagenstos, The Structural Turn in AntidiscriminationLaw, 94 CAL. L. REV. 1 (2006); Olatunde C. A. Johnson, Disparity Rules, 107 COLUM. L. REV. 374 (2007); Gillian E. Metzger, Administrative Constitutionalism,91 TEX. L. REV. 1897 (2013); Charles F. Sabel & William H. Simon, Contextualizing Regimes: Institutionalization as a Response to the Limits of Interpretationand Policy Engineering, 110 MICH. L. REV. 1265, 1285-91 (2012); Susan Sturm, Second GenerationEmployment Discrimination:A StructuralApproach, 101 COLUM. L. REV. 458 (2001). Yale Journal on Regulation Vol. 33, 2016 they cannot confidently promulgate or apply substantive rules to remedy problems that generate such claims. Thus, they have been drawn to an alternative approach: lawmakers can require the institutional actors to assess their own conduct, then appraise the adequacy of this self-assessment. This regulatory approach has an affinity with the core techniques of post-bureaucratic organization, which are designed to reduce the behavioral unreflectiveness and normative ambiguity that create problems for classic civil rights doctrine. Classic civil rights doctrine tends to treat managerial inquiry and control as prerequisites for responsibility; the emerging duty treats them as entailments of responsibility. The reform of policing exemplifies this evolution. Post-bureaucratic transformation came late to policing, but its manifestations are now pervasive. Courts have been a major influence. This influence has not been transmitted primarily through declarations of substantive rights enforced through the exclusionary rule or damage actions. The most important avenue of judicial influence in recent years has instead been structural reform. In many cases where a private plaintiff or the Department of Justice has alleged recurring civil rights violations, intervention has focused on changes in administrative processes. Appellate authority, however, remains conflicted about such intervention, as some judges urge deference to administrative discretion for fear that structural relief would rigidify administration. In doing so, they often appear to assume mistakenly that such intervention would have to take bureaucratic forms. Although the trend toward post-bureaucratic reform is clear, we note two ambiguities. First, post-bureaucratic policing can take different organizational forms. In particular, alternatives vary in the extent to which they emphasize innovation and decentralization. Second, judicial remedies differ in the extent to which they focus on specialized procedures for civil rights compliance, as opposed to broader reforms that reach into the agencies' core activities. We illustrate the contrasting trajectories reform might take through a comparison of New York, where a federal district court held policing practices unlawful in 2013, and Cincinnati, which settled a civil rights challenge to policing practices in 2002. Both examples manifest the structural turn. They owe more to an implicit duty of responsible administration than to any particular substantive norm. They tend to mandate the key elements of post-bureaucratic administration-explicit but provisional policy-setting on matters previously left to tacit discretion, monitoring, and reassessment in light of experience and evidence. However, the two regimes embody the opposing poles of post-bureaucratic policing. New York's, sometimes called Assertive Policing (AP), focuses on rapid deployment of personnel to implement a limited set of standard solutions, especially street confrontations and minor-offense enforcement. By contrast, Cincinnati has adopted an approach called Problem-Oriented Policing (POP) that emphasizes varied, innovative, and localized responses, often developed in Police Accountability collaboration with stakeholders. It thus typifies a form of public initiative known as democratic experimentalism.4 The two cities also reflect different approaches to judicial remediation. The New York intervention emphasizes specialized procedures designed to constrain civil rights violations. Cincinnati's intervention, on the other hand, required comprehensive reform of the city's policing practices, in particular, the adoption of POP. The scope of the Cincinnati intervention is unique among judicially- induced resolutions in policing cases. Some provisions that appear increasingly in settlements, however, blur the distinction between specialized and systemic reform by requiring reassessment of crime-control tactics associated with recurrent civil rights violations. The more ambitious of these reforms resemble holistic civil rights interventions in other areas, including labor standards, education, and child welfare. In Part II, we show that the changing nature of civil rights claims has pushed doctrine to focus on administration, but that the move has been intermittent and incomplete. In Part II1, we discuss the evolution of policing. We show that policing has evolved beyond the bureaucratic forms assumed by classic civil rights doctrine, but that this evolution involves multiple trajectories with potentially different implications for civil rights enforcement. In Part IV, we contrast the conventional, judicially-supervised reform in New York with the more ambitious initiative in Cincinnati and suggest some advantages of the latter. Existing research does not establish the superiority of either model (in part because it often fails to distinguish them). Yet the Cincinnati approach has potential advantages for both crime control and civil rights that warrant experimentation and research. In particular, it appears less prone than the New York approach to antagonize and deter cooperation from minority communities and better able to take account of the costs of indiscriminate criminalization of nonviolent disorderly conduct.

11. The Evolution of Civil Rights Doctrine

Confronted by new problems that resist substantive regulation, civil rights doctrine has increasingly addressed administration. It has imposed duties that require defendants to clarify and assess rigorously their own interpretations of the norms that govern them. The trend, however, has been halting, and doctrine sometimes perpetuates older premises about organization that are, in important respects, anachronistic.

4. Michael C. Dorf & Charles F. Sabel, A Constitution of Democratic Experimentalism, 98 COLUM. L. REV. 267, 288 (1998). Yale Journal on Regulation Vol. 33, 2016

A. The OrganizationalPremises of Classic Doctrine

Classic doctrine drew on two models from the past. The first, which dates from the early years of the republic, sees public officials as autonomous actors 5 exercising broad discretion within fairly clear, judicially elaborated constraints. The second, which dates from the Progressive and New Deal eras, sees them as bureaucrats exercising low-visibility discretion in the interstices of webs of hierarchically promulgated rules.6 Doctrine often simply ignored the organizational context of government and treated officials as lone individuals. When it did recognize public organization, it tended to ignore managerial action other than making or following rules, and it more or less explicitly disregarded managerial inaction. When deciding whether to intervene, the courts sometimes acted or spoke as if their intervention would necessarily take a quasi-bureaucratic, rule-based form. They treated the decision to intervene as a choice between the judicial imposition of rules or deference to administrative discretion, and often decided to hold back for fear of excessively cramping discretion. The tendency to see government as operating either through independent individual action or bureaucracy is salient in the two core substantive civil rights doctrines that address frontline policing-antidiscrimination and search and seizure. It can also be seen in procedural doctrine on injunctive relief and on attribution of frontline conduct to agencies or senior managers.

1. Antidiscrimination.

The premise of the autonomous official is most salient in antidiscrimination doctrine. Liability turns here on "intent" to discriminate. Application of the idea was fairly straightforward when lawsuits challenged rules that explicitly distinguished among races or genders or practices that officials discussed in explicitly racist or sexist terms. However, partly as a result of the success of past litigation, explicitly racist or sexist rules and official discourse have virtually disappeared from public life. "Second-generation" challenges typically address decisions or practices that are not facially discriminatory but that foreseeably or demonstrably harm protected groups disproportionately. If an employer makes hiring decisions under legitimate but vague standards like "diligent" and "resourceful," bias may not be evident in any one decision. If the overall pattern disfavors a protected group, however, suspicion arises. An employer's use of a specific rule, like a high school graduation requirement, also generates concern

5. JERRY L. MASHAW, CREATING THE ADMINISTRATIVE CONSTITUTION: THE LOST ONE HUNDRED YEARS OF ADMINISTRATIVE LAW 65-78 (2012). 6. William H. Simon, The Organizational Premises of Administrative Low, 78 LAW & CONTEMP. PROBS. 61, 67-79 (2015). On the idea of bureaucracy as a combination of rigid rules and low- visibility discretion, see MICHEL CROZIER, THE BUREAUCRATIC PHENOMENON (1964), and ALVIN W. GOULDNER, PATTERNS OF INDUSTRIAL BUREAUCRACY (1954). Police Accountability where the rule, though facially neutral, disqualifies more black than white candidates. However, discriminatory intent is more elusive in a world of tacit discrimination. Even when the challenged action is motivated by group-based animus, it may be hard to prove the animus when the defendant's agents take care to hide it. More fundamentally, once we get beyond overt prejudice, it is hard to define, much less discover, intent. The Supreme Court says that the challenged decision must have been made "because of, not merely in spite of' the harm it inflicts on a protected group. 7 This seems ambiguous or seriously under-inclusive. In organizations, harmful conduct may often lack purpose and reflection. It can arise from "selective indifference," cognitive stereotyping, or inertial perpetuation of routine.8 In such cases, what is objectionable is precisely the actor's inattentiveness to the harm. Doctrine has responded by allowing plaintiffs to support their cases with evidence of disparate outcomes or effects.9 If hiring decisions under general standards disproportionately disfavor women or the high school diploma requirement disproportionately disadvantages blacks, the courts sometimes recognize a rebuttable inference of discrimination. The inference has to be rebuttable because there are possible legitimate explanations for the disparities. Perhaps high school graduation reliably predicts better job performance. The key question is the strength of the burden of rebuttal. If a facially non-frivolous recitation of a legitimate purpose is enough, much unfairness will not be redressed. On the other hand, requiring the defendant to produce rigorous scientific validation for its decisions may generate overbroad liability because such validation is either prohibitively expensive or inconclusive. The courts have been especially sensitive to the dangers of constraining legitimate practice through excessive liability in the criminal justice context. In declining to entertain a challenge to sentencing practices based on exceptionally rigorous disparate impact evidence, the Supreme Court said in McCleskey v. Kemp that giving weight to such evidence "throws ... into serious question the

7. Pers. Adm'r of Mass. v. Feeney, 442 U.S. 256, 279 (1979). See generally Sturm, supra note 3, and Johnson, supra note 3. 8. JOHN DEWEY, HUMAN NATURE AND CONDUCT (1922) (on inertial perpetuation of routine); Paul Brest, In Defense of the Antidiscrimination Principle, 90 HARv. L. REV. 1, 14 (1976) (on "selective indifference"); Linda Krieger, The Content of Our Categories:A Cognitive Bias Approach to Discrimination and Equal Employment Opportunity, 47 STAN. L. REV. 1161 (1995) (on cognitive stereotyping). On the general problem of "racism without racists," see RICHARD THOMPSON FORD, THE RACE CARD 37-92 (2008). 9. Pamela L. Perry, Two Faces of DisparateImpact Discrimination,59 FORDHAM L. REV. 523, 581-91 (1991). Disparate impact liability has been explicitly recognized under Titles VI (federal grantees) and VII (employers) of the Civil Rights Act of 1964, Title VIII (housing marketers) of the Civil Rights Act of 1968, and under the Age Discrimination in Employment Act. Texas Dep't of Hous. and Cmty. Affairs v. Inclusive Communities Project, Inc., 135 S.Ct. 2507, 2516-19 (2015); Johnson, supra note 3, at 386-401. Yale Journal on Regulation Vol. 33, 2016 principles that underlie our entire criminal justice system."' Such deference, however, leaves less overt or visible forms of abuse immune from accountability.

2. Search and Seizure.

The second view of organizational liability in classic doctrine-the view premised upon the twentieth-century bureaucratic model-is salient in search and seizure doctrine. This doctrine rejects the subjective "intent" focus of antidiscrimination. Instead, its touchstone is "objective reasonableness."'"I One might have thought that this perspective would lead in policing, as it did in common law professional negligence, to broad supervision under norms derived from professional culture and practice. Fourth Amendment reasonableness, however, differs from the common law duty of care in negligence actions. It is a set of more or less specific norms promulgated by the courts (or occasionally, legislatures) on the basis of an ostensibly utilitarian calculus. 2 The courts insist that these norms take the form of "readily administrable rules." They emphasize that Fourth Amendment norms have to be "applied on the spur of (and in the heat of) the moment" and thus cannot contain too many "ifs, ands, or buts." 3 Indeed, even as the courts try to make the rules as simple as possible, they do not hold officers accountable for unlawful practices unless the courts' prior pronouncements unambiguously cover the situation at hand. Thus, the "qualified immunity" doctrine provides that liability for unreasonable searches and seizures can only be imposed where the action violates a "clearly established" duty. 14 In effect, the courts treat frontline officers like low-level bureaucrats. And they cast themselves in the role of bureaucratic rulemaker. At the same time, they recognize the importance of broad frontline discretion, which they sometimes treat as effectively unregulatable. When a court finds that the plaintiffs' claim cannot be formulated as an administrable substantive rule, it dismisses the claim.'5

10. 481 U.S. 279, 315 (1987). See also id. at 282 ("Because discretion is essential to the criminal justice process, we would demand exceptionally clear proof before we would infer that discretion has been abused .... "). 11. Graham v. Connor, 490 U.S. 386, 396 (1989). 12. John Rappaport, Second Order Regulation ofLaw Enforcement, 103 CAL. L. REV. 205, 215-17 (2015) (reviewing Fourth Amendment and other cases in which courts have imposed specified conduct rules on law enforcement). 13. Atwater v. City of Lago Vista, 532 U.S. 318, 347 (2001). 14. Brosseau v. Haugent, 543 U.S. 194, 199 (2004) (per curiam) (holding that officers are not liable for "constitutionally deficient conduct" unless the deficiency was "clearly established" in a "particularized sense" relating to the circumstances of the officer's challenged act). Some cases even take the view that only judicial authority within the circuit where the conduct occurred can clearly establish a duty. E.g., Thomas ex rel. Thomas v. Roberts, 323 F.3d 950, 955 (11th Cir. 2003). But other cases disagree. E.g., Owens v. Lott, 372 F.3d 267,279-80 (4th Cir. 2004). 15. Atwater, 532 U.S. at 350 (stating as ground for rejection that "plaintiff's proposed rule... promise[s] very little in the way ofadministrability"). Police Accountability

3. Attribution Under Section 1983.

Both autonomous-individual and rule-based perspectives underlie doctrine on the attribution of frontline conduct to public institutions. Individualism is salient in the practice of naming individual officers as defendants and, sometimes, only individual officers. Naming individual officers is partly a formalistic evasion of the traditional sovereign immunity of the federal government and the states. But even where doctrine permits suing the government by name-for example, in the case of municipalities-plaintiffs purport to seek relief against individuals, despite the fact that the officers are virtually always indemnified for liability. With both public entity and senior officer defendants, the question arises when such defendants are accountable for the wrongdoing of frontline officers. dealt with this question extensively under 42 U.S.C. § The Supreme Court has 16 1983-the procedural vehicle for most civil rights suits against state officials. Early in the development of 1983 doctrine, the courts rejected importing the private law principle of respondeat superior. Doing so would have made public agencies (or their heads) in effect strictly liable for most wrongful subordinate conduct intended to advance the agencies' public purposes. Respondeat superior seemed to risk too much judicial intrusion. The courts could have responded to this problem by predicating entity or employer liability on a showing of irresponsible-that is, negligent or reckless-mismanagement or failure to manage subordinates. Instead, at least initially, they demanded a showing that the agent's conduct was, in some sense, "authorized." The conventional form of authorization was a "policy," which was most readily demonstrated by a hierarchically promulgated rule.1 7 In the landmark Monell case, the Supreme Court rejected the claim that the "mere right to control without any direction or control having been exercised" was sufficient.' 8

4. Structural Relief.

Finally, organizational premises surfaced in the ambivalence toward structural relief in classic doctrine. The presumptive forms of relief for police wrongdoing were motions to suppress illegally seized evidence and damage judgments. Both procedures had well-recognized limits. Suppression was only available in the small fraction of police-citizen encounters that resulted in the filing of charges. Damage actions required large investments of energy and resources. When these remedies did prove effective for complainants, they often

16. 42 U.S.C. § 1983 (2012) (providing for a cause of action against anyone who, "under color of' state law, causes a deprivation of a federal right). 17. HAROLD S. LEWIS, JR. & ELIZABETH J. NORMAN, CIVIL RIGHTS LAW AND PRACTICE 193-201 (2d ed. 2004). 18. Monell v. Dep't of Soc. Servs., 436 U.S. 658, 694 n.58 (1978) (citing Rizzo v. Goode, 423 U.S. 362, 370-71 (1976)). Yale Journal on Regulation Vol. 33, 2016

involved what seemed to be excessive public costs, especially where suppression thwarted otherwise valid prosecutions. Moreover, responsibility for challenged conduct was usually diffuse and ambiguous, and officers almost never bore liability costs personally. Neither procedure, therefore, had demonstrably strong deterrent effects. Nevertheless, these remedies made sense from the perspective of bureaucratic organization. Bureaucracies acted systemically through rules. If a rule was bad, declaratory judicial relief could correct it. Individual frontline wrongdoing, however, was assumed to be idiosyncratic.' 9 Case-by-case remediation of the sort provided by suppression motions or damage actions was 20 well designed to correct idiosyncratic error. Of course, the courts in the classic era recognized some situations where reactive and individualized intervention was plainly inadequate. Beginning with schools and moving to other public institutions, they developed the structural injunction. This form of relief became highly controversial, and the appellate courts became ambivalent about its use. They never repudiated it, but they issued various cautions to the lower courts. For reasons of respect for other levels or branches of government or of relative expertise, appellate doctrine has portrayed structural intervention as a last resort. The strictures have been especially severe with respect to policing, the subject of two landmark cases disapproving structural challenges. In Rizzo v. Goode, the Supreme Court reversed an order mandating that the Philadelphia police adopt a complaint process consistent with "generally recognized minimum standards."21 The original order was based on evidence of sixteen incidents of frontline misconduct over the course of a year. The court noted that there is no independent "right" to an adequate complaint process. It ruled that the evidence of instances of misconduct did not suffice for systemic relief where the conduct was not authorized by departmental policy and senior 22 officials "played no affirmative part" in it. In City of Los Angeles v. Lyons, the Court reversed an order enjoining the use of "chokeholds" in certain circumstances and mandating training programs and record-keeping to insure compliance with the prohibition. 23 There was no dispute in Lyons that at least some of the challenged conduct was systemic because it was authorized by department policy. But the Court reversed for lack of standing. The lone plaintiff, it held, could not assert the threat of "real and immediate" injury based on a single past encounter in which he had been

19. See JAMES P. WOMACK ET AL., THE MACHINE THAT CHANGED THE WORLD: THE STORY OF LEAN PRODUCTION 57 (1990) ("In [bureaucratic] plants, problems tended to be treated as random events. The idea was simply to repair each error and hope that it didn't recur."). 20. See generally William Stuntz, The Political Constitution of Criminal Justice, 119 Harv. L. Rev. 780, 818-30 (2006). 21. Rizzo, 423 U.S. at 370. 22. Id. at 377. 23. City of Los Angeles v. Lyons, 461 U.S. 95, 100, 102 (1983). Police Accountability improperly subjected to a chokehold. 24 The Court emphasized that the possibility that he would be subjected to it again appeared small and speculative. The Court has not always refused to recognize standing on the basis of a small probability of official injury.25 The case therefore seems to reflect, in part, deference to police discretion. The underlying premise of much classical doctrine is that managerial inquiry and control are prerequisites of duty rather than entailments of it. Although the premise is pervasive, it is hard to find an explicit defense of it. It appears to rest on an assumption that organizations take the form of a classical bureaucracy in which senior officers influence conduct only by commanding it through rules. It follows that they are not responsible for conduct they have not mandated by rule (i.e., "authorized"). Perhaps the courts also believe that there are no standards by which they could define affirmative duties of responsible 2 6 administration apart from the commands of substantive law.

The doctrine thus ignores that organizations in recent decades have been less prone to take bureaucratic forms. Moreover, even when organizations are formally bureaucracies, it is well recognized that senior managers influence frontline practice in ways other than through promulgating rules. They can selectively fail to enforce rules. They can make resources available for some practices and not others, or they can measure and reward some conduct while ignoring other conduct. Indeed, to limit accountability of senior officials to violations they know about and/or authorize is to leave doctrine powerless against one of the most characteristic pathologies of modem organizations- strategically selective knowledge and attention. Managers monitor and enforce the goals they care about, while ignoring how their subordinates achieve their performance levels. Corporate executives can set and reward large sales targets without paying attention to frauds or kickbacks. Likewise, police executives can set and reward targets for stops and arrests without paying attention to Fourth Amendment violations. Managers may feel they are worse off if they have knowledge about compliance with norms that impede their primary goals. ' '27 Ignorance gives them "deniability.

24. Id.at 102. 25. See Clapper v. Amnesty Int'l USA, 130 S.Ct.1 138, 1155-65 (2013) (Breyer, J., dissenting) (discussing several cases allowing standing on the basis of speculative prospect of injury from official action and arguing that they are indistinguishable from the instant case in which the majority denied standing, citing Lyons, to a challenge to national security surveillance practices). 26. Even in the private sphere, the duty of corporate managers to manage proactively was not clearly recognized until the 1990s. See In re Caremark Int'l Derivative Litig., 698 A.2d 959,969- 70 (Del. Ch. 1996). 27. See, e.g., Craig Haney & Donald Specter, Treatment Rights in Uncertain Times, in TREATING ADULT AND JUVENILE OFFENDERS WITH SPECIAL NEEDS 51, 70 (Joseph B. Ashford et al. eds., 2001) (reporting testimony by the head of the California Department of Corrections and Rehabilitation Yale Journal on Regulation Vol. 33, 2016

B. The Emerging Duty ofResponsible Administration

Public administration has evolved away from the bureaucratic model presupposed in the classic cases. This evolution has influenced civil rights doctrine, but only fitfully and unevenly. Bureaucracy lends itself to situations where there is confidence in relatively stable and uniform interventions. In situations where problems and solutions are not well understood or where intervention has to take account of varying contexts, bureaucracy is less effective. Demand for government to respond to social volatility and diversity has grown in recent decades. Accordingly, administration has moved toward a style that can be called post-bureaucratic. Post-bureaucratic organization does not focus on balancing stable rules and lightly supervised discretion.28 Its central mandate to senior managers is not rule promulgation but planning, monitoring, and reassessment. Plans differ from rules in being more comprehensive and more provisional. Monitoring is important not only to induce compliance with the dictates of the plan but also to facilitate learning. A key part of the manager's job is to collect and publicize information about unanticipated problems and successes so that frontline agents can learn from each other and the agency can learn from peer institutions engaged with

that he resisted screening inmates for mental illness "because he knew that once mentally ill individuals were identifiable he would be responsible for treating them"). On the role of "deliberate ignorance" in contemporary political and business misconduct, see William H. Simon, Wrongs of Ignorance and Ambiguity: Lawyer Responsibilityfor Collective Misconduct, 22 YALE J. ON REG. 1, 3-9 (2005). 28. Our contrast between bureaucratic and post-bureaucratic organization is based on a vast literature observing and recommending a basic transition in organizational form. Although there are varying formulations, the contrast can be presented usefully as two ideal types: Bureaucracy,in the mid- twentieth century conception, is a balance of rules and low-visibility discretion. The basic idea is to implement a program developed at the top and revised only episodically. Frontline discretion thus tends to be regretted and minimized. Nevertheless, because full compliance is thought unattainable and perhaps also undesirable, a residuum of such discretion is accepted. This residuum is unavoidable because monitoring capacity is limited; it is also potentially benign to the extent that it enables frontline workers to mitigate harshness or waste in situations where application of the rules would be counter-productive to their underlying purposes. Three structural features follow: (1) the paradigmatic norm is the rule. Rules tend to be inflexible and to be interpreted formally; (2) monitoring of frontline agents focuses on compliance with the rules, but because it is expensive and demoralizing, monitoring is limited and reactive, focused especially on responding to complaints; and (3) rules tend to be stable, revised only episodically and in processes centered at the top. What we call post-bureaucraticorganization rejects both inflexible rules and low-visibility discretion. Senior officials view program norms as provisional and expect to develop them in light of experience gained at the frontline. Organization tries to combine continuous improvement with transparency and frontline initiative with accountability. The characteristic structural features are: (1) the paradigmatic norm is the plan. Plans are more comprehensive than rules, and their norms are interpreted purposively. Frontline agents are expected to depart from such norms when following them would be counter-productive, but agents must signal their departures in ways that trigger review of their decisions; (2) monitoring is proactive and based on audits as well as complaints. Monitors assess not just compliance with the norms but also the effectiveness of the practice prescribed by the norms; and (3) norms are revised more or less continuously in the light of information from monitoring. Frontline workers participate in the process of norm revision. For discussion and citations to the literature, see Charles F. Sabel, A Real Time Revolution in Routines, in THE CORPORATION AS A COLLABORATIVE COMMUNITY 106-56 (Charles Hecksher & Paul Adler eds., 2006); Simon, OrganizationalPremises, supra note 6. Popular accounts include PETER SENGE, THE FIFTH DISCIPLINE: THE ART AND PRACTICE OF THE LEARNING ORGANIZATION (1990), and WOMACK ET AL., supranote 19. Police Accountability comparable problems. Reassessment involves deliberative engagement between and among senior managers, frontline agents, and, where appropriate, stakeholders, about the ends and means of intervention. Such engagement fuels continuous rearticulation of the plan. The rule of law implications of post-bureaucratic administration are different from those expressed in classic civil rights doctrine. Post-bureaucratic administration insists on self-consciousness and explicitness. Where it finds unreflectiveness and ambiguity, it sees them not as intractable conditions of organizational life but as symptoms of administrative failure. Where such failure manifestly threatens civil rights values, judges can intervene without becoming bureaucrats themselves. They can require administrators to make policies explicit, to give reasons for them, to supervise their implementation in a transparent way, and to reassess periodically. We should not expect public officials to have broad discretion over the degree to which they will be accountable for their exercise of discretion. Inducing post-bureaucratic reflection and transparency makes practice more predictable to citizens and facilitates political mechanisms of oversight. Transparent administrative practice makes it easier for courts to apply whatever substantive constraints there are on practice. Moreover, when practice is reliably articulated across jurisdictions, both courts and political agencies may be able to derive minimum substantive standards empirically. They can do so by determining which practices have widespread acceptance and put pressure on outliers to adopt them (unless they can produce good explanations for not doing so).29 This post-bureaucratic structural approach has been incorporated into some important civil rights statutes and regulations. Instead of categorically defining prohibited conduct, these laws mandate that actors make plans to vindicate a value or achieve a goal, monitor the implementation of the plan, and reassess the plan in light of experience. Examples include the provisions of the Juvenile Justice and Delinquency Prevention Act requiring that states plan to reduce disproportionate minority contacts in the criminal justice system30 and those of the Prison Rape Elimination Act requiring that prison officials plan to achieve "a zero-tolerance standard" for sexual assault.3'

29. Compare Tenn. v. Gamer, 471 U.S. 1, 18-19 (1985) (invoking the articulated practice standards in most police agencies as support for holding that deadly force cannot be used against fleeing suspected non-violent felons), with Whren v. United States 517 U.S. 806, 814-15 (1996) (dismissing the claim that Fourth Amendment reasonableness should be measured by "usual police practices" or the conduct of a "reasonable [police] officer" and saying that the Court could not "plumb the collective consciousness of law enforcement"). Pertinent here is the argument that many fundamental public law principles develop through a process ofdeliberative engagement and experimentation that leads to the identification and judicial condemnation of outliers. E.g., WILLIAM N. ESKRIDGE, JR. & JOHN FEREJOHN, A REPUBLIC OF STATUTES (2010). 30. Johnson, supra note 3, at 402; 42 U.S.C. § 5633(a)(22) (2000) (requiring state efforts to "reduce the disproportionate number of juvenile members of minority groups who come into contact with the juvenile justice system"). 31. Prison Rape Elimination Act of 2003 § 3, 42 U.S.C. § 15602 (2012). Yale Journal on Regulation Vol. 33, 2016

The "reasonable accommodation" requirement for employers in the Americans with Disabilities Act is another important development. 32 Like the Juvenile Justice statute, it abandons the intent requirement of classic doctrine and requires reasonable proactive assessment and mitigation of disparate impacts. Unlike the Juvenile Justice Act and Prison Rape Act, it does not specifically mandate planning, monitoring, and reassessment. However, it gives employers incentives to engage in such activities in order to demonstrate compliance. Judicial doctrine also reflects these trends, but it remains constrained by the organizational premises of classicism. Some decisions have moved toward the structural approach. 33 Yet, the courts still sometimes speak as if their only options were to impose rigid rules on the one hand or to defer to unaccountable discretion on the other. We find both progress and constraint in the key areas that bear on policing-antidiscrimination, search and seizure, 1983 attribution, and systemic relief.

1. Antidiscrimination.

Commenters have argued that the best way for the courts to apply general antidiscrimination norms to second-generation problems is to recognize an affirmative duty to make reasonable efforts to investigate, assess, and mitigate disparate harms to protected groups.34 Reasonableness would then imply the kinds of post-bureaucratic procedures specifically mandated in statutes like the Juvenile Justice Act and Prison Rape Act. No cases have followed this path explicitly, but some have done so indirectly. The indirect approach treats reasonable proactive efforts as a rebuttal to inferences drawn from evidence of harm to protected groups. For example, when plaintiffs produce evidence of disparate impact under Titles VI, VII, or VIII of the Civil Rights Act, defendants must produce evidence of a business rationale for the decisions. Although the authority varies on the strength of this burden, it clearly requires more than a recitation of a legitimate purpose. Employers sometimes produce elaborate, methodologically rigorous studies that make explicit the criteria on which decisions are based and validate the predictive value of these criteria for productivity. The courts sometimes suggest that even demonstrably predictive criteria are unacceptable if there are

32. 42 U.S.C. § 12111 (2012). See also 42 U.S.C. § 12143(c)(7) (2012) (provision of the Americans with Disabilities Act mandating that certain public transportation systems develop plans to accommodate disabled passengers). 33. Rappaport, supra note 12, at 220-31, reviews cases in several constitutional areas- including reapportionment, school desegregation, voting rights, and employment discrimination-taking a structural ("second-order") approach to rights elaboration. 34. Krieger, supra note 8, at 1245 (proposing that Title VII be interpreted to create a "prescriptive duty to identify and control for errors in social perception and judgment which inevitably occur, even among the well-intended"); Richard Thompson Ford, Bias in the Air: Rethinking Employment Discrimination Law, 66 STAN. L. REv. 1381, 1384 (2014) (arguing for a "duty of care to avoid unnecessarily perpetuating social segregation or hierarchy"). Police Accountability equally effective (or perhaps, almost as effective) alternatives that are less harmful to the protected group. (Perhaps a college degree predicts productivity, but so would an honorable discharge from the military). In theory, the purpose of the stronger rebuttal requirements is to negate the inference that the asserted purpose is a "pretext" for purposeful discrimination. However, rebuttal is often expensive, and it is unusual to demand this amount of substantiation for a party's denial of wrongdoing. A better explanation for requiring an employer to critically examine practices that disproportionately disadvantage protected groups is that, given the stakes for the group members and the social commitment to equality, it would be irresponsible not to examine them. Moreover, cases holding that a demonstrably valid criterion is insufficient when there are less harmful alternatives, even if the less harmful ones are slightly more expensive to administer, seem to interpret the general non-discrimination language of the Civil Rights Act to imply something like the "reasonable accommodation" requirement of the Americans with Disabilities Act. Pamela Perry notes that while disparate impact doctrine usually purports to follow an "intent theory," the more demanding cases are better understood in terms of a "fault" theory that presupposes a duty to take reasonable care to avoid disparate impacts. 35 At this point, the duty of non-discrimination has become in substantial part a duty of responsible administration. In criminal justice, however, the courts have tended to resist disparate impact evidence and insist on direct proof of intent with respect to individual discrimination claims.3 6 As discussed below, class claims for systemic relief are another matter.

35. Perry, supra note 9, at 581-91. See also Christine Jolls,Antidiscrimination and Accommodation, 115 HARV. L. REV. 642, 651-683 (2001) (demonstrating that many cases decided under authority that does not explicitly require reasonable accommodation impose liability for the defendant's failure to mitigate harm to protected groups even though mitigation is costly). 36. See Wayte v. United States, 470 U.S. 598, 610 (1985) (holding that a showing that a prosecutorial policy had a "discriminatory effect" was insufficient and that a challenger must show "that the government intended such a result"); United States v. Armstrong, 517 U.S. 456, 469-70 (1996) (holding that a litigant alleging selective prosecution must plead specifically that similarly situated people were treated differently before pursuing discovery); Angela Davis, Prosecution and Race: The Power and Privilege of Discretion, 67 FOROHAM L. REV. 13, 31 (1998) (asserting that Armstrong makes challenges to selective prosecution a "virtual impossibility"); Samuel Gross & Katherine Y. Barnes, Road Work: Racial Profilingand Drug Interdiction on the Highway, 101 MicH. L. REV. 651, 741 (2002) (stating that no case has approved suppression on the basis of statistical proof). Also pertinent here is Whren v. United States, 517 U.S. 806 (1996), which permits "pretextual" police stops by holding that the Fourth Amendment requirement of probable cause does not require that there be such cause for the suspicion that motivated the stop as long as there is cause for suspicion of some crime, for example, a minor traffic violation. Whren was formally a Fourth Amendment case, but the plaintiff specifically argued for judicial regulation on the ground that pretextual stops facilitated race discrimination. Some believe that, given the difficulty of proving discriminatory intent, Whren "conferred upon police virtual carte blanche to stop people because of the color of their skin." I WAYNE R. LAFAVE, SEARCH AND SEIZURE 1.4(e), at 122 (3d ed. 1996). Yale Journal on Regulation Vol. 33, 2016

2. Search and Seizure.

Of the four doctrines we are considering, substantive search and seizure doctrine has evolved the least, though its limitations are increasingly recognized.3 7 The dominant perspective in substantive Fourth Amendment jurisprudence has been the autonomous officer. The courts most often assess the "objective reasonableness" of challenged practice from the point of view of the individual officer at the point where she decides whether to intervene.38 Courts have recognized a duty on the part of that officer to make reasonable efforts to inform herself within the confines of the situation. 39 Where the inference that prompted the initial stop is "dispelled by information gained" in the course of the stop, she must forego further detention or search.4° But neither the officer nor the department is accountable for her state of knowledge prior to the encounter. This is important because this ex ante knowledge is not independent of the department's practices. It is a function of the policing styles and structures the department chooses. A department that invests in gathering intelligence and making it available to frontline officers may stop people who turn out to be law- abiding less frequently than one that does not. Yet, at least under substantive doctrine, the reasonableness inquiry does not extend to the agency's background efforts to develop information. The court has emphasized the narrowness of the range within which the reasonableness norm operates. In Whren v. United States, the Supreme Court said that Fourth Amendment regulatory efforts were largely for "searches or seizures conducted in an extraordinary manner," such as those involving deadly force, entry into dwellings, or bodily invasion.4 Whren involved a more routine "pretextual" search in which the police used a traffic violation, for which there was probable cause, as an excuse for a search motivated by suspicion of a more serious crime. After holding that motive was irrelevant to "objective reasonableness," the Court went on to state that any probable cause was sufficient to establish reasonableness. The Court recognized that enforcement of traffic laws is massively under-inclusive but denied that Fourth Amendment reasonableness imposed any constraint on decisions as to what searches and seizures to conduct among those for which there is probable cause.42 In another

37. E.g., Rappaport, supra note 12, at 231-64 (arguing for more emphasis on second- order judicial regulation under the Fourth Amendment). 38. Graham v. Connor, 490 U.S. 386, 396 (1989). 39. E.g., United States v. Brugal, 185 F.3d 205, 210 (4th Cir. 1999) (holding a search unreasonable where defendant explained his initially suspicious highway exit by saying that he needed gas and the officer could have verified by examining gas gauge), rev'den banc, 209 F.3d 353 (4th Cir. 2000). 40. Brugal, 185 F.3d at 210. See also Terry v. Ohio, 392 U.S. 1, 30 (1968) (defining the Fourth Amendment standard for a police stop as the "reasonable suspicion" of an individual officer). 41. Whren, 517U.S. at818. 42. Whren, 517 U.S. at 816-19. Police Accountability case, the Court specifically rejected the suggestion that there should be any obligation to adopt the "less restrictive alternative" among the available 43 enforcement options. Under current doctrine, then, the Fourth Amendment reasonableness norm does not regulate the agency's efforts to develop information or its enforcement strategy. The persistent bureaucratic conception of organization explains this limitation. For the court, constraint must take the form of a more or less 44 categorical rule. When such rules are infeasible, the court must withdraw. Again, however, the story becomes more complicated when we look at attribution doctrine and remedial practice in class actions.

3. Attribution.

Doctrine has moved considerably on the issue of when institutions will be chargeable for the wrongful conduct of subordinate agents. The most dramatic development has been in the private sector under Title VII. In Faragherv. City of Boca Raton, the Supreme Court adopted an approach designed to avoid both strict liability and judicial withdrawal. This was a sexual harassment case in which the plaintiff proved many instances of indisputable misconduct on the part of middle- and lower-level employees. The Court rejected the defendant's claim that liability required "active or affirmative, as opposed to passive or implicit, misuse of supervisory authority."45 Instead, it held that such workplace misconduct would be presumptively attributed to the employer. However, the employer could rebut this presumption by showing "that it had exercised reasonable care to avoid harassment and to eliminate it when it might occur, and that the complaining employee had failed to exercise like reasonable care."4 6 Faragher appears to have prompted pervasive corporate efforts to develop and monitor sexual misconduct policies. In the public sector, courts in 1983 cases have qualified the classical insistence on top-level authorization by holding that "deliberate indifference" on the part of senior administrators will suffice. Like "discriminatory intent," deliberate indifference is a concept that owes more to doctrinal desperation than

43. Atwater v. City of Lago Vista, 532 U.S. 318, 350-51 (2001) ("The logic of such elaborate less-restrictive-alternative arguments could raise insuperable barriers to the exercise of virtually all search-and-seizure powers.") (quoting United States v. Martinez-Fuerte, 428 U.S. 543, 557-58 n.12 (1976)). 44. Whren, 517 U.S. at818-19 ("[W]e are aware ofno principle that would allowusto decide at what point ... infraction itself can no longer be the ordinary measure of the lawfulness of enforcement."). 45. Faragher v. City of Boca Raton, 524 U.S. 775, 804 (1998). 46. Faragher,524 U.S. at 805. See also Burlington Indus. v. Ellerth, 524 U.S. 742, 765 (1998) (recognizing as part of an affirmative defense to a sexual harassment claim that "the employer exercised reasonable care to prevent and correct promptly" harassing behavior). For interpretations that see such cases as a more general trend, see Bagenstos, supra note 3, at 15-20; Ford, supra note 34, at 1403-15; and Sturm, supra note 3, at 479-521. Yale Journal on Regulation Vol. 33, 2016

psychological insight. In practice, it is established by passivity in the face of knowledge of subordinate misconduct. In addition, courts have recognized 1983 liability for inadequate training, supervision, and screening of employees. 47 In some respects, the doctrine parallels the enforcement duties recognized in the Title VII employment context in Faragher.However, the 1983 authority has yet 48 to specifically recognize duties to promulgate policies and acquire information. Perhaps the duty to "supervise" is broad enough to encompass such activities. If so, it would be a small step in principle from this authority to a duty of responsible administration. Some commentators, however, are pessimistic about 49 the practical prospects for such expansion. The rejection of respondeat superior in favor of "deliberate indifference" for the purposes of attribution has committed the courts to make some judgments about the adequacy of administrative effort. Doing so requires them to broaden the individualistic perspective they often adopted in substantive discrimination and search and seizure doctrine. Substantive doctrine takes the local perspective of the frontline officer in a particular situation either subjectively (with discrimination) or objectively (with search and seizure). Attribution requires that we step back and examine some of the factors that determined how he got there.

4. Structural Relief.

While the Supreme Court has on occasion cautioned lower courts against excessive zeal, it has not categorically denied the legitimacy of systemic relief, and the lower courts have given such relief against a broad range of public authorities. Rizzo and Lyons did not end structural relief against police departments. In a few cases, Rizzo has been distinguished by proof of a larger number of instances of unlawful conduct or by evidence of explicit or implicit managerial approval or encouragement.5" Lyons has not been applied to several racial profiling claims,

47. Bd. ofCnty. Comm'rs v. Brown, 520 U.S. 397, 419 (1997); City of Oklahoma v. Tuttle, 471 U.S. 808, 813 (1985); Walker v. City of New York, 974 F.2d 293, 297-98 (1992). 48. But see Thomas v. Cook Cnty. Sheriffs Dep't, 588 F.3d 445, 454 (7th Cir. 2009) (finding as evidence that plaintiff's injuries arose from official policy "the failure to have a system in place to allow prompt review of inmates' medical requests, the practice of severely under-staffing correctional officers, and the failure to fix broken video monitors" (emphasis added)). 49. See Barbara Armacost, OrganizationalCulture and Police Misconduct, 72 GEO. WASH. L. REV. 453,486-90 (2004) (asserting that "failure to train cases are notoriously difficult to litigate and even more difficult to win" and discussing a remarkable example). 50. E.g., Thomas v. Cnty. of Los Angeles, 978 F.2d 504 (9th Cir. 1992) (noting that after Rizzo, "[slpecific findings of a persistent pattern of misconduct supported by a fully defined record can support broad injunctive relief"); Easyriders Freedom F.I.G.H.T. v. Hannigan, 92 F.3d 1486, 1501 (9th Cir. 1996) (upholding a permanent injunction against the California Highway Patrol's citation policy, citing Thomas, where the citations received by the plaintiffs were "the result of a clear CHP citation policy in violation of the Fourth Amendment"); Pennsylvania v. Porter, 659 F.2d 306, 322 (3d Cir. 1981) (upholding injunctive relief against a police officer, mayor, and chief of police where the police chief, "unlike the city officials in Rizzo v. Goode, . . . had knowledge and was intimately involved" in encouraging the police officer's violation of civil rights). Police Accountability which can be understood to involve a diffuse stigmatic injury to an entire group, 5 1 and many systemic search and seizure claims have a racial dimension. Moreover, standing is not a problem for the federal government. In 1994, in the aftermath of the Rodney King trial and the ensuing riots, Congress authorized the federal government to seek relief against a "pattern or practice" of police 52 conduct that violated federal law. In any event, most cases settle. No doubt these settlements reflect some feeling by defendants that they are vulnerable under the substantive law. However, some, including Cincinnati and New York, have been influenced by political forces catalyzed by the suit. And while defendants usually resent the continuing "outside" scrutiny that settlements require, much of what the settlements prescribe involves practices that many peer departments have adopted voluntarily and are widely considered to be good practice within the professional community. Peer acceptance may make the provisions more palatable to some defendants. The fact that cases settle for reasons not entirely dependent on the substantive merits opens space for negotiated remedies to depart from the technical eccentricities of doctrine. Although there is substantial variation among remedies, some best-practice norms seem to be emerging. In particular, about twenty agreements concluded by the Department of Justice have had wide-ranging influence outside of the cases in which they were negotiated. According to Samuel Walker and Carol Archbold, they have defined "the core elements of [a] new police accountability."5 3 Walker and Archbold call the dominant pattern PTSR, for Policy, Training, Supervision, and Review. 54 It includes requirements that the agency promulgate detailed standards, train staff in the standards, and monitor compliance with them. (A major part of the plaintiff's evidence in the recent New York case was based on records that the defendant was required to keep under a consent decree in an earlier case).55 The core of the new monitoring regime consists of (1) civilian complaint procedures; (2) use-of-force or critical-incident policies that require investigations wherever police use or threaten physical force; and (3) early intervention systems that use civilian complaints and use-of-force reports

51. Brandon Garrett, Standing While Black: Distinguishing Lyons in Racial Profiling Cases, 100 COLUM. L. REV. 1815 (2000). 52. 42 U.S.C. § 14141 (2012). 53. SAMUEL E. WALKER & CAROL A. ARCHBOLD, ThE NEW WORLD OF POLICE ACCOUNTABILITY 192 (2014). For a detailed account of a notable decree with a favorable appraisal of its success, see CHRISTOPHER STONE ET AL., POLICING LOS ANGELES UNDER A CONSENT DECREE: THE DYNAMICS OF CHANGE AT THE LAPD (2009). 54. WALKER & ARCHBOLD, supranote 53, at 16; see also Armacost, supra note 49, at 528-31 (discussing DOJ consent decree practice). See Note, Complex Enforcement: Unconstitutional Prison Conditions,94 HARv. L. REV. 626,638-40 (1981) (noting converging remedial provisions in prison cases: "while they are not constitutional rights as such, they seem to represent the criteria of legality and therefor are more than mere remedies"). 55. Floyd v. City of New York, 959 F. Supp.2d 540, 609 (S.D.N.Y. 2013) (citing Daniels v. City of New York, No. 99 Civ. 1695 (S.D.N.Y. Sept. 24, 2003)). Yale Journal on Regulation Vol. 33, 2016 to detect patterns of poor performance by individual officers or their supervisors and intervene with warnings, training, or discipline. While the settlements typically describe substantive policies minimally or vaguely, they often go into detail about procedure. For example, the Oakland, California consent decree specifies twenty indicators to be tracked in the department's early 56 intervention system. The complaint, early intervention, and use-of-force processes generate signals designed to alert managers to problems. At a minimum, managers should consider whether the signals suggest a need for individual training, counseling, or discipline. More ambitiously, review may consider systemic implications. Department of Justice standards provide that use-of-force assessments should "include an examination of the police tactics and the precipitating events that led to the use of force" and consideration of whether the incident "suggests the advisability of revising or reformulating agency policy, strategy, tactics, or training." 57 Although limited to use-of-force review, such measures move the agency in the direction of the continuous and systemic reassessment demanded by post-bureaucratic organization. They treat error not as idiosyncratic, but as potentially symptomatic of broader dysfunction.

III. The Evolution of Policing

The key organizational assumptions of classic civil rights doctrine-that managers control subordinate conduct mainly through rules and that minimally accountable frontline discretion is inevitable-fit well with both the ideology and practice of mid-twentieth century policing. However, policing has changed more than doctrine. The dominant policing models are post-bureaucratic. Their implications for civil rights remediation are ambiguous because post- bureaucratic administration can take different forms. We illustrate the range of possibilities with two cases: AP in New York and POP in Cincinnati.

A. From Reaction to Proaction

Policing was one of the key cases for mid-twentieth century sociologists seeking to revise the idea of bureaucracy to acknowledge that the top-down rules emphasized in Max Weber's conception 58 virtually always co-existed with low- visibility frontline discretion. Tacit discretion could take a malign form as

56. WALKER & ARCHBOLD, supra note 53, at 148. 57. U.S. DEP'T OF JUST., PRINCIPLES FOR PROMOTING POLICE INTEGRITY: EXAMPLES OF PROMISING POLICE PRACTICES AND POLICIES 5 (2001). For an example, see POLICE ASSESSMENT RESOURCE CTR., THE PORTLAND POLICE BUREAU: OFFICER-INVOLVED SHOOTINGS AND IN-CUSTODY DEATHS 143-60 (2003). 58. 1 MAX WEBER, ECONOMY AND SOCIETY 212-301 (Guenther Roth & Claus Wittich eds., Ephraim Fischoffet al. trans., 1978). Police Accountability

arbitrariness or corruption or a benign form as contextual adaptation, but it was considered in one form or another unavoidable.59 In this mid-century view, which was shared by both academics and officers, policing was dominantly reactive. Police responded to calls for service and reports of crimes. Their key measures of success were response time for the former and case closure rates for the latter. Police also patrolled, but patrol tended to be undirected or directed in terms of broad "sector-and-shift" categories. As James Q. Wilson reported, "[F]ew police administrators show much interest in 'planning' the deployment of their manpower and equipment. There is no information-and in the nature of the case, there can never be sufficient information-on the effects of alternative police strategies on the several kinds 60 of crime. Policing in this view was also incident-based. The basic unit of analysis was a threatened or completed breach of law or public order. Incidents were self- contained. Departments credited officers with success where they prevented a harm, sanctioned a completed wrong, or mediated a dispute to the satisfaction of those involved. Their interventions were limited to traditional law-enforcement strategies-interrogations, arrests, warnings, or guidance about legal requirements. Internal control and accountability were weak, and external control was highly limited. The courts held police accountable to civil rights norms in the cases that reached them, but these represented only a very small fraction of police activity. In principle, electoral control of local government held police accountable, but the political levers-appointment of top-level officials and budgetary control-were crude, and voters and civilian officials had limited information. The most salient indicators-aggregate measures of crime and disorder-were thought only weakly correlated with police efficacy. Beginning in the 1980s, the mid-century view was gradually repudiated in favor of a post-bureaucratic view.61 Inthis latter view, stable top-down rules are supplemented, and in some respects displaced, by more flexible norms-notably, plans and indicators. Plans are more comprehensive but more provisional than rules, and indicators measure results rather than dictate practice. The basic unit of analysis in the new view is not the incident but the pattern or the problem. Incidents that claim the attention of the police tend to recur at a particular location, harm a recurring victim, and/or involve a recurring perpetrator. These incidents often have common causes that call for coordinated responses. Proactive organizations "map"crime incidents to determine where and how they should concentrate their efforts.

59. A classic expression of this view is JAMES Q. WILSON, VARIETIES OF POLICE BEHAVIOR: THE MANAGEMENT OF LAW AND ORDER IN EIGHT AMERICAN COMMUNITIES (1968). See also HERMAN GOLDSTEIN, PROBLEM-ORIENTED POLICING 5-30 (1990) (characterizing the view critically). 60. WILSON, supra note 59, at 60. 61. See WALKER & ARCHBOLD, supra note 53, at 1-56. Yale Journal on Regulation Vol. 33, 2016

The new view breaks with prior assumptions about discretion. While the old view treated discretion as a residual, barely licit category, the new one explicitly encourages it. At the same time, it insists that discretion be accountable. Decisions and strategies are reviewed both before and after the fact through various procedures by supervisors, peers, and stakeholders.

B. Two Trajectories of Reform: Assertive Policing vs. Problem-Oriented Policing

As described so far, the proactive view has become a consensus. But at this level of generality, it has basic ambiguities. Recent discussion of policing simmers with new concepts. Two contrasting ideal types best illustrate the range of variation: AP and POP. Both are the subjects of extensive literatures. 62 We focus in particular on experiences with AP in New York from 1993 through 2013, and with POP in Cincinnati from 2001 to the present. We do not offer a comprehensive account of either regime, but our contrast captures differences in tone and emphasis. Neither city has implemented any single model fully and both regimes contain elements from both of our ideal types. Moreover, both AP and POP are controversial. New York has observed remarkable reductions in crime over more than two decades, but the causes of these reductions have been unclear, and the city's style of policing provoked massive political opposition for its effects on racial minorities. A federal district court partially enjoined the department's "stop-and-frisk" practice in 2013, and Bill de Blasio partially repudiated the practice during his mayoral campaign and upon assuming office in 2014.63 Cincinnati has seen substantial crime reductions in recent years, but again, the causes remain to be demonstrated. However, there is evidence that police relations with minority communities have improved, and 64 there have been some notable local crime-control successes.

62. For discussions of AP, see KENNEDY SCH. OF GOV'T, ASSERTIVE POLICING, PLUMMETING CRIME: THE NYPD TAKES ON CRIME IN NEW YORK CITY (1999); JACK MAPLE & CHRIS MITCHELL, THE CRIME FIGHTER: PUTTING THE BAD GUYS OUT OF BUSINESS (1999); ELI SILVERMAN, NYPD BATrLES CRIME: INNOVATIVE STRATEGIES IN POLICING (1999); and Dennis C. Smith & William J. Bratton, PerformanceManagement in New York City: COMPSTA T and the Revolution in Police Management, in QUICKER, BETTER, CHEAPER?: MANAGING PERFORMANCE IN AMERICAN GOVERNMENT 453-82 (Dail W. Forsythe ed., 2001). For discussions of POP, see HERMAN J. GOLDSTEIN, PROBLEM-ORIENTED POLICING (1990); and MICHAEL S. SCOTT & STUART KIRBY, CTR. FOR PROBLEM ORIENTED POLICING, IMPLEMENTING POP: LEADING, STRUCTURING, AND MANAGING A PROBLEM-ORIENTED POLICE AGENCY (2012). 63. Floyd v. City of New York, 959 F. Supp. 2d 540, 667 (S.D.N.Y. 2013). For de Blasio's campaign position, see Safe Streets, Safe Neighborhoods Across New York City, BILLDEBLASIO.COM, http://www.billdeblasio.com/issues/crime-fighting-public-safety (last visited Sept. 21, 2015) (on file with Yale Journalon Regulation); for his position as mayor, see Benjamin Weiser & Joseph Goldstein, Mayor Says New York City Will Settle Suits on Stop-and-Frisk Tactics, N.Y. TIMES, Jan. 30, 2014, at Al. Bratton left as New York chief in late 1994. He subsequently served as chief in Los Angeles and was then re-appointed chief in New York in 2013. His thinking had evolved since his departure, and on return, he distanced himself from the regime in place. 64. CINCINNATI POLICE DEP'T ET AL., COLLABORATIVE AGREEMENT ANNUAL PROBLEM SOLVING REPORT (2006) [hereinafter 2006 PROBLEM SOLVING REPORT] (reporting some local Police Accountability

In broad summary, the AP model assumes the efficacy of standard interventions, especially stops and arrests. It uses data on crime incidence to rapidly deploy resources to "hot spots" and to hold officers down the chain of command accountable for rapid responses to crime indicators. The regime de- emphasizes rules and induces some initiative at the precinct command level. However, like the bureaucracy against which it reacts, it remains a principal- agent model of action. It assumes that the principal or senior official can confidently know what needs to be done and that the chief organizational problem is inducing subordinate agents to execute the plan. By contrast, POP assumes that standard responses are typically ineffective, even when efficiently directed to high-crime areas. It looks for a broader array of patterns than AP, aspires to analyze them more deeply, and customizes solutions. Like AP, POP maps spatially, but it analyzes in more detail and views crime occurrences as evidence of environmental and social conditions that facilitate crime. It then tries to devise interventions that disrupt or reconfigure these conditions. POP also employs a form of pattern analysis that focuses on violent people as well as places. When it identifies persistent offenders, it responds with a package of threats, offers of social services, and moral 65 exhortation tailored to the specific circumstances of the actors. POP draws on knowledge and encourages initiative from both frontline officers and community members. While AP tends to emphasize the lines between supervisors and beat officers and between the police and the 66 community, POP tends to blur them.

successes); ROBIN S. ENGEL & M. MURAT OZER, CINCINNATI POLICE DEPARTMENT 2014 CRIME SUMMARY: A DECADE IN REVIEW (2015) (on file with authors) (reporting a 40.5 percent reduction in violent crimes, a 33.8 percent reduction in citizen complaints, and a 64.5 percent reduction in use-of-force incidents from 2004 to 2014); GREG RIDGEWAY ET AL., POLICE-COMMUNITY RELATIONS IN CINCINNATI (2009) (reporting improvements in popular perceptions of police following court-supervised reforms involving problem-oriented policing); Robin S. Engel et al., Reducing Gang Violence Using Focused Deterrence:Evaluating the Cincinnati Initiative to Reduce Violence (CIRV), JUST. Q. 1. 28-32 (2011) (methodologically sophisticated study finding that Cincinnati's anti-violence initiative has been effective). Cincinnati's police force is much smaller than New York's (991 officers versus 34,555), and has considerably fewer police personnel per capita (37.6 per 10,000 versus 59.5 per 10,000). Police Employment, Officers Per Capita Rates for U.S. Cities, GOVERNING.COM, http://www.goveming.com/gov-data/safety-justice/police-offcers-per-capita-rates-employment-for-city- departments.html (last visited Oct. 31, 2015). Cincinnati has a significantly higher poverty rate (31.3 versus 20.9). U.S. City Poverty Rankings, SYRACUSE.COM, http://www.syracuse.com/news/index.ssf/20l4/09/uscitypovertyrankings look up compare-nations biggest cities.html (last visited Oct. 31, 2015). 65. This approach to violence is often referred to as "focused deterrence" or "pulling levers policing," but as one of its originators notes, it is best considered an elaboration of POP. David M. Kennedy, Old Wine in New Bottles: Policingand the Lessons of Pulling Levers, in POLICE INNOVATION 160 (David Weisburd & Anthony Braga eds., 2006). 66. AP resembles what some reformers call "new public management." See Christopher Hood, The "New Public Management " in the 1980s: Variationson a Theme, 20 ACCT. ORGS. & SOC'Y 93 (1995). POP resembles what we call "experimentalism." See Dorf& Sabel, supra note 4. Neither term seems common in the policing field. Yale Journal on Regulation Vol. 33, 2016

1. Assertive Policing

a. Strategy.

Since 1993, when William Bratton became Chief of Police for the first time, New York has, in both practice and self-presentation, emphasized AP. The basic strategy was to identify and then take control of crime-infested locales "by street stops of suspicious-looking persons, by frisking after stops for weapons or contraband, and by making arrests for minor offenses as a way to remove perceived risks from the street and to identify persons wanted for other crimes. ' 67 The strategy was initiated with major investments in Compstat-information technology that enables prompt identification of geographic patterns, or "hot spots." Commanders are expected to deploy patrol officers promptly to these locations and then to redeploy them as data indicate changes in crime incidence. The strategy was summarized by Jack Maple, Bratton's principal deputy at the time the regime was established, as "cops on dots." Maple and others called Chief of Patrol Louis Anemone "our Patton," invoking the World War I general 68 associated with mobile tank warfare. The visible presence of police at a hot spot might reduce crime, but AP did not rely only on presence. It prescribed confronting and searching people who appeared to be engaged in illegal activity, and arresting or citing people for offenses either observed by the officers or discovered when suspects were stopped and searched. Police occasionally observed serious offenses, and searches occasionally discovered unlicensed guns. But most arrests and summonses were for minor offenses; the largest category involved marijuana use. The regime designers saw such activity as deterring serious crime for various reasons. Although it was not part of the official explanation, evidence at the federal trial suggested that some officers thought that aggressively confronting young men would instill general fear that inhibited criminal activity.69 The "broken windows" theory formulated by George Kelling and James Q. Wilson suggested that "quality-of-life" policing could prevent the emergence of hot spots in transitional neighborhoods by encouraging law-abiding people to act as crime-inhibiting "eyes on the street" and to provide information to the police.70 Bratton and Maple favored minor-offense enforcement in high-crime neighborhoods because it enabled the police to put pressure on people they believed, but could not prove, were engaged in more serious offenses. Prosecution for minor offenses might temporarily incapacitate such people; it

67. FRANKLIN E. ZIMRING, THE CITY THAT BECAME SAFE: NEW YORK'S LESSONS FOR URBAN CRIME CONTROL 118 (2012). 68. KENNEDY SCH., supra note 62, at 17-21; MAPLE & MITCHELL, supra note 62, at 31, 120; Smith & Bratton, supra note 62, at 457-62. 69. Floyd v. City of New York, 959 F. Supp. 2d 540, 606 (S.D.N.Y. 2013). 70. George Kelling & James Q. Wilson, Broken Windows: The Police and Neighborhood Safety, THE ATLANTIC (March 1982). Police Accountability might lead to more intensive surveillance through probation; or it might induce 71 them to provide intelligence about the criminal activities of others. The regime tended to assume that, within a hot spot, criminal activity was diffuse, interconnected, and relatively homogeneous. Thus, it focused on broadly targeted and uniform interventions. It considered the appropriate response, more often than not, to be an increase in stops, searches, and arrests. From January 2004 to June 2012, the NYPD documented 4.4 million street stops. At its peak in 2011, it made 686,000. Half of the stops were followed by a search for weapons, and twelve percent led to arrests or summons, most for minor offenses. 72 For the department, a major performance indicator-perhaps the major performance indicator-has been the quantity of stops and related enforcement activity. The city denied that there have ever been enforcement quotas for precincts or officers either before or after a 2010 ordinance forbidding them. But evidence at trial, including recordings of precinct meetings and surveys of officers, indicated that commanders and officers felt strong pressure. Moreover, a senior police manager testifying at trial acknowledged that the number of stops was one factor in performance assessment. His main qualification was that the department also considered the "quality" of stops, which he defined repeatedly as "one that's in the right place, the right time, for 73 the right crime. The designers emphasized motivation, rather than innovation, as the key shortcoming in the prior regime. 74 Decades of complacent management and poor morale had left officers timid and indolent. They needed to be pushed. Thus, reforming managers used their control over promotions to reward the conduct they approved, and they used meetings with borough and precinct commanders to publicly honor and shame in accordance with their views of performance.75 A

71. KENNEDY SC[., supranote 62, at 9-13; MAPLE & MITCHELL, supra note 62, at 153- 56; ZIMRING, supra note 67, at 117-31. 72. Floyd, 959 F. Supp. 2d at 573-75. 73. Transcript of Testimony by Joseph Esposito, Chief of Dep't, New York City Police Dep't, at 2983, Floyd v. City of New York, 959 F. Supp. 2d 540 (S.D.N.Y. 2013) (No. 08 Civ. 1034) [hereinafter "Transcript of Esposito Testimony"]; Floyd, 959 F. Supp. 2d at 592-607. Following the court's ruling, William Bratton, having resumed the role of chief, acknowledged that under his predecessor officers were "pushed hard" to increase stops and frisks. Pervaiz Shallwani & Sean Gardiner, NYPD Officers "Pushed" on Stop-and-Frisk: Police Commissioner Bratton, WALL ST. J. METROPOLIS BLOG (Jan. 30, 2014, 2:43 PM), http://blogs.wsj.com/metropolis/2014/01/30/nypd-officers-pushed-on-stop-and- frisk-police-commissioner-bratton. 74. See Transcript of Esposito Testimony, supra note 73, at 3020, 3039 (discussing problems of motivating underperforming officers); MAPLE & MITCHELL, supra note 62, at 7 (characterizing the majority of officers on most forces as lazy or indolent). 75. "One officer [in Lowell, Massachusetts] described Compstat as a forum where officers had their 'balls ripped off and surmised that this only served to make individuals 'reluctant to speak up ... reluctant to do their job."' JAMES J. WILLIS ET AL., COMPSTAT AND ORGANIZATIONAL CHANGE IN THE LOWELL POLICE DEPARTMENT 15 (2003). Some assert that such pressures have led to cheating on reports. See, e.g., JOHN ETERNO & ELI SILVERMAN, THE CRIME NUMBERS GAME: MANAGEMENT BY MANIPULATION (2012). Yale Journal on Regulation Vol. 33, 2016 precinct commander won approval by showing mastery of current data about crime patterns and by quickly deploying officers in response. The effect of AP on centralization is complex. The key focus of the regime designers was on the precinct commanders. They were subject to more intensive scrutiny from the center, but since this scrutiny was focused on results, it left them discretion with respect to tactics. As we have noted, however, commanders had less discretion in practice because of the pressure to show immediate response to crime rate increases and the presumption that the appropriate response was to stop-search-arrest. It appears that commanders did not encourage initiative on the part of frontline officers. 76 Bratton has stated that creativity should not be expected from patrol officers, who tend to be 77 inexperienced and untrained in the relevant skills. Finally, AP rejects more ambitious versions of "community policing," a philosophy that emphasizes development of deep local knowledge and active engagement with local leaders, residents, and business owners. Mayor Rudolph Giuliani, who brought on Bratton to inaugurate Compstat, had dismissed such a model as "social work," a term that for him connoted timidity and sentimentality. Bratton had a more developed critique. In addition to his skepticism about the capacities of patrol officers, he doubted that community activists were meaningfully representative of their communities or had much information to contribute that could not be gathered through conventional investigatory or data mining techniques. The original Kelling-Wilson "broken windows" idea emphasized the contribution of law-abiding residents to crime control through informal pressures. However, Bratton's and Maple's reinterpretation of it saw minor-offense enforcement mainly as leverage for the police over serious 78 wrongdoers, who were assumed to be diffused throughout the community.

b. Limitations.

Despite the phenomenal declines in major crimes in New York, reservations about AP have become prominent. Some concern limitations of its tactics as instruments of crime control; others concern insensitivity to collateral costs. First, the preoccupation with static efficiency-moving police to hot spots-led to unreflective reliance on a narrow set of interventions. Officers were not encouraged to innovate, and indeed, the emphasis on immediate response may have inhibited impulses to do so. The most detailed account to date of

76. A survey by the Police Foundation of Compstat nationwide found that most frontline officers were not familiar with Compstat data and were "rarely called upon to explain a particular decision." David Weisburd et al., Changing Everything So That Everything Can Remain the Same: Compstat and American Policing,in POLICE INNOVATION, supra note 65, at 291. 77. KENNEDY SCH., supranote 62, at 9-10. 78. See KENNEDY SCH., supra note 62, at 9-10; Smith & Bratton, supra note 62, at 466- 72. See also HUMAN RIGHTS WATCH, A RED HERRING: MARIJUANA ARRESTEES Do NOT BECOME VIOLENT FELONS 14-15 (2012) (quoting officials and commentators asserting that marijuana arrests contribute to declines in serious crime). Police Accountability

Compstat implementation-a study of the Lowell, Massachusetts program modeled on New York's-reports, "Compstat's data orientation did seem to affect ... when and where responses would be mobilized, but it had generally done little to stimulate analysis of how to actually respond on the basis of the 79 data." Performance measures may have been too coarse for meaningful assessment of practice. Lacking the ability to compare different interventions within the city and uninterested in efforts elsewhere, the city tended to measure efficacy solely in terms of crime rate declines. AP espoused the principle of '80 "relentless follow-up. However, in practice, that notion appears to have meant, at best, observation of whether crime rates declined following intervention, and, at worst, observation to confirm that stop-and-frisk practices were being implemented without any regard to their efficacy. Although crime declines were dramatic, their relation to AP practices is ambiguous. Crime rates do not seem to have been responsive to any fine-grained measure of changes in enforcement practice. Trends do not seem to have been strongly affected either by the dramatic increase in stop-and-frisk activity from 2004 to 2011, or its dramatic 8 decrease in 2012 (probably in response to the lawsuit and political protest). ' Moreover, the department's claims for the efficacy of stop-and-frisk omit many relevant variables, including:

[A] significant increase in the New York City police force, a general shift in drug use from crack cocaine to heroin, new computerized tracking systems that speed up police response to crime, favorable economic conditions in the 1990s, a dip in the number of eighteen to twenty-four- year-old males, an increase in the number of offenders currently incarcerated in city jails and state prisons, the arrest82 of several big drug gangs in New York, as well as possible changes in adolescent behavior.

Second, the Compstat metrics took little or no account of major costs of the stop-and-frisk practices. Three sorts of costs now seem especially important. First, there are the costs of unlawful stops-stops that do not meet constitutional

79. WILLIS ET AL., supra note 75, at 48 (emphasis omitted). The Police Foundation study of Compstat implementation nationwide found that "[t]he vast majority of problem-solving approaches identified in these model Compstat agencies relied on traditional police strategies that had been used before ..." David Weisburd et al., Changing Everything, supra note 65, at 292. 80. MAPLE & MITCHELL, supra note 62, at 32. 81. See Philip Bump, New York Has Essentially EliminatedStop-and-Frisk - and Crime Is Still Down, WASH. POST, Dec. 3, 2014, https://www.washingtonpost.com/news/the- fix/wp/2014/12/03/new-york-has-essentially-eliminated-stop-and-frisk-and-crime-is-still-down. 82. BERNARD HARCOURT, ILLUSION OF ORDER: THE FALSE PROMISE OF BROKEN WINDOWS POLICING 94 (2001). See also ZIMRING, supranote 67, at 131-50, who finds that the practice of quickly deploying officers to "hot spots" probably had a significant impact but, like Harcourt, finds little evidence for the efficacy of minor-offense enforcement. Studies of"hot spot policing" often blur the distinction between AP and POP, both of which emphasize deployment on the basis of spatial patterns. However, a recent meta-analysis that takes some account of differences in strategic frameworks concludes that "problem-oriented policing interventions generate larger mean effect sizes when compared to interventions that simply increase levels of traditional police actions in crime hot spots." Anthony A. Braga et al., The Effects of Hot Spots Policingon Crime: An UpdatedSystemic Review and Meta-A nalysis, 31 JUST. Q. 633, 656 (2014). Yale Journal on Regulation Vol. 33, 2016 standards.83 The federal district court emphasized the city's failure to make serious effort to monitor these costs. Second, there are the costs of stops that are lawful (because there was reasonable suspicion) but fail to produce evidence of unlawful activity.8 4 Most of these stops cause at least inconvenience and often anxiety and humiliation to law-abiding people. And third, there are the costs of minor-offense enforcement to the people charged and to their families and communities.85 The legal status of the second and third categories of costs is ambiguous, but they have come to be viewed as important by a large fraction of New Yorkers. Stops and minor-offense enforcement, even on otherwise adequate grounds, are especially resented because they are disproportionately directed at minority groups. Many now assert that the costs of minor-offense enforcement are especially large. 86 Most of post-stop enforcement action involved offenses such as marijuana or alcohol consumption, trespass, or non-threatening forms of disorderly conduct. These offenses are not regarded as serious in themselves. But each enforcement action creates a record that increases the likelihood that the subject will receive subsequent and harsher attention from the criminal justice system. Many create public records that will impair the subject's employment and housing prospects. A strategy of policing minor misconduct in high-crime neighborhoods, even if implemented solely on the basis of non-racial indications of misconduct, will disproportionately affect minorities because they live disproportionately in high-crime areas. Minority neighborhoods may benefit from reduced crime, but they will suffer to the extent that law-abiding residents find the life chances of their friends and family members cumulatively impaired by repeated police encounters. The department ignores these costs. Indeed, it has treated arrests and summonses even for minor offenses as measures of success.

83. The Floyd court found on the basis of a study of records by Jeffrey Fagan that "at least 200,000 stops [out of 4.4 million between 2004 and 2012] were made without reasonable suspicion" and that the "actual number ... was likely far higher." Floyd v. City of New York, 959 F. Supp. 2d 540, 559-60 (S.D.N.Y. 2013). 84. About 88 percent of the stops between 2004 and 2012 did not lead to any law enforcement action. Id. at 953. The remaining 12 percent resulted in arrests or "summonses" (for minor violations that do not warrant taking the person into custody), but it seems likely that many of these people were not engaged in unlawful activity. In some years, nearly half of the charges were dismissed or adjourned in contemplation of dismissal. N.Y. STATE OFFICE OF THE AVr'y GEN., A REPORT ON ARRESTS ARISING FROM THE NEW YORK CITY POLICE DEPARTMENT'S STOP-AND-FRISK PRACTICES 8-9, 22-23 (2013) (reporting data from 2009 through 2012). 85. "[O]nly 1.5 percent of stops between 2009 and 2012 resulted in a jail or prison sentence of any duration ..... Id. at 10. 86. See MICHELLE ALEXANDER, THE NEW JIM CROW: MASS INCARCERATION IN THE AGE OF COLORBLINDNESS (2010). Police Accountability

2. Problem-Oriented Policing

a. Strategy

Although the idea of POP antedates AP, no jurisdiction has implemented POP with the degree of ambition and comprehensiveness with which New York implemented AP. The problem-oriented approach is promoted and supported by the U.S. Department of Justice and a national Center for Problem-Oriented Policing. Several empirical studies have shown at least modest benefits from problem-oriented initiatives.87 The approach has been embraced by many departments, but rarely as a basis for reorganizing its core operations. And many initiatives have proved fragile, falling victim to budget cutbacks and senior 8 8 management turnover. Cincinnati adopted POP in 2002 as part of a "Collaborative Agreement" settling one of two lawsuits challenging its use-of-force practices as unconstitutionally arbitrary and discriminatory. The implementation process was rocky,89 but when the agreement terminated in 2008, the court-appointed monitor reported that the problem-oriented approach had been strongly institutionalized. Some ground was lost subsequently due to fiscal cutbacks and managerial turnover, but commitment to POP survived and has strengthened in recent years, substantially due to the efforts of a group of district commanders who have had favorable experiences with it. On a brief visit in 2014, we found officers at many levels to be articulate and enthusiastic about POP.9"

87. DAVID WEISBURD ET AL., CAMPBELL COLLABORATION, THE EFFECTS OF PROBLEM-ORIENTED POLICING ON CRIME AND DISORDER (2008) (meta-analysis of POP studies concluding that they show modest gains); Braga et al., supra note 82, at 24 (meta-analysis of "hot spots" policing studies concluding that they show significant gains in the aggregate and generally larger gains for problem-oriented interventions); Anthony A. Braga & Brenda J. Bond, Policing Crime and Disorder Hot Spots: A Randomized Controlled Trial, 46 AM. J. CRIMINOLOGY 577, 595 (2008) (describing an experiment in Lowell, Massachusetts, in which POP-inspired treatment interventions achieved significantly better effects than conventional practice at control sites and in which success at treatment sites was due to "situational strategies" rather than minor offense enforcement). 88. One city where at least moderately robust implementation has been documented is Lowell, Massachusetts. After a disappointing experiment with New York-style Compstat, see WILLIS ET AL., supra note 75, the City undertook POP initiatives with extensive monitoring and analysis. Reports indicate that the efforts have become increasingly sophisticated and have been generally effective. BRENDA J. BOND ET AL., U.S. DEP'T OF JUST., LOWELL, MASSACHUSETTS, SMART POLICING INITIATIVE: REDUCING PROPERTY CRIME INTARGETED HOT SPOTS (2014) (describing a subsequent initiative in which more sophisticated POP efforts outperformed control sites); Braga & Bond, supra note 87, at 578 (2008) (describing a 2005-06 experiment in which "shallow" POP strategies outperformed conventional practice). 89. Among the problems: one of the two organizational plaintiffs withdrew from the settlement after adopting a more confrontational stance toward the City. The Community Partnering Center-created to coordinate community involvement in problem-solving--dissolved when some leaders opted for aggressive protest tactics at the expense of problem-solving efforts. Shortly after the settlement, gang-related violence spiked and prompted the city to adopt AP-type stop-and-frisk practices for a time. 90. SAUL GREEN, MONITOR'S TRANSITION YEAR REPORT ON THE COLLABORATIVE AGREEMENT BETWEEN THE PLAINTIFFS AND THE CITY OF CINCINNATI (2008); John Eck & Jay Rothman, Police Community Conflict and Crime Prevention in Cincinnati, Ohio, in PUBLIC SECURITY AND POLICE Yale Journal on Regulation Vol. 33, 2016

Cincinnati, like most cities, shows some influence of Compstat. The district commanders meet with the chief and senior staff weekly and review current crime data. They use the Compstat crime mapping techniques. 9' Yet, many officers distinguish their approach from New York's. In the first place, there is a strong rhetorical influence on innovation and creativity. The Collaborative Agreement that mandated POP proclaimed as its core tenet that "problems are dilemmas to be engaged and learned from."92 In particular, there is an explicit repudiation of the idea that confrontation and arrests should be the presumptive responses: "A law enforcement response is always a possibility, but may not be required. Rather, a range of options is explored . . . ."" On our visit, we frequently heard officers say, "We couldn't arrest our way out of this problem." Examples of problems for which arrest is thought usually ineffective are street prostitution and retail drug markets. As long as environmental conditions remain unaltered, new recruits will take the place of those arrested or those arrested will return when released. Cincinnati POP proponents reject aggressive, indiscriminate "zero tolerance" or "broken windows" enforcement. The department responded to a spike in murders in 2006 with Operation Vortex, which included practices resembling New York's stop-and-frisk ones, but abandoned them the following year. Lieutenant Colonel James Whalen, who commanded Operation Vortex, told a reporter that he had concluded that indiscriminately confrontational approaches were "bullshit:" "Even in high crime neighborhoods, there are a lot

REFORM IN THE AMERICAS 225-37 (John Bailey & Lucia Dammert eds., 2006). We have also relied on Ted Wojcik, Problem-Oriented Policing in Cincinnati: An Update (May 5, 2015) (on file with authors), which is based on interviews conducted in the spring of 2015. The robustness of POP, despite turmoil in the upper reaches of the department and its political surroundings, is explained in significant measure by the institutional spaces that the settlement afforded younger officers attracted to innovative police responses and the career opportunities it opened to those who proved adept at developing them. Several in the cohort of officers who came of age under the settlement are now in or rising to senior positions and provide critical support for POP. In addition, at least one segment of the Cincinnati regime has been strongly institutionalized. This is the gang-focused intervention called the Cincinnati Initiative to Reduce Violence (CIRV).When inaugurated, it produced short-term success, but the effects were not sustained- a common experience with initiatives of this kind. The City concluded that implementation had suffered from overdependence on specific personnel and informal relations. It thus undertook a thoroughgoing reorganization, drawing on management practices devised at Proctor and Gamble, the city's major corporate citizen. The resulting structure has an executive director who reports to a committee consisting of the mayor and two other city-wide officials. Three multi-agency task forces formulate strategies and revise them continuously on the basis of data produced by a monitoring committee. Marie Tillyer et al., Beyond Boston: Applying Theory to Understandand Address SustainabilityIssues in FocusedDeterrence Initiativesfor Violence Reduction, 58 CRIME & DELINQUENCY 973 (2012) (analyzing the re-organization of CIRV). 91. An example of a project that combines creativity with Compstat-style static efficiency concerns involves the overlay of geographical data on traffic citations with accident and crime data to enable the deployment of traffic enforcement resources to locations where it is most likely to prevent harm. Daniel W. Gerard, Cincinnati HAZARD: A Place-Based Traffic Enforcement and Violent Crime Reduction Strategy, THE POLICE CHIEF, Jul. 2013, at 44, 44-46. 92. Collaborative Agreement at 2, In re Cincinnati Policing, No. 1-99-317 (S.D. Ohio Aug. 5, 2002) [hereinafter "Collaborative Agreement"]. 93. ld. at 6. Police Accountability of honest people living there. Meanwhile, the real bad guys-they know a sweep is on, so they stay inside until things cool off."94 Where arrests are part of a problem-oriented strategy, they are used as a last resort and applied in as precisely targeted a way as possible. In pointing to data supporting the effectiveness of the Cincinnati Initiative to Reduce Violence (CIRV), the initiative that succeeded Operation Vortex, Captain (then Lieutenant) Mars Herald said, "[T]he results are particularly impressive because they were 95 achieved with so few arrests." The intellectual core of problem-oriented practice is a discipline specifically mandated by the Collaborative Agreement called SARA-Scanning, Analysis, Response, Assessment. It begins with a precise definition of a problem, proceeds to look for well-configured interventions, implements them, assesses the results, and then, if the problem persists, begins the cycle anew with a revised account of the problem in light of experience. POP emphasizes the potential complexity of both problems and interventions. Illustrations are given in a series of more than seventy problem- specific guides produced by the national Center for Problem-Oriented Policing (e.g., "Assaults in and Around Bars," "Disorder at Day Laborer Sites," and "Shoplifting"). The guide on "Street Prostitution" explains that the "problem" associated with prostitution could be the exploitation of the prostitutes by their pimps, harm to minors, sexually-transmitted disease, the impact of street solicitation on neighborhood atmosphere, or exploitation of customers (notably, by robbery). Each interpretation implies a different set of interventions, and the problem often turns out to have many facets and require interventions with 96 multiple prongs. In general, problems tend to emerge in two sorts of patterns. One involves recurring criminal incidents at a common location. The other involves recurring 97 lawlessness by a single person or group. Locational analysis in POP involves a thicker sort of mapping than in Compstat. It includes considerations of social influence and economic interdependence, as well as geographical incidence.9 One set of strategies for

94. John Seabrook, Don't Shoot: A Radical Approach to the Problem of Gang Violence, THE NEW YORKER, June 22, 2009, at 32. His successor as Vortex commander, Captain Daniel Gerard, expressed a similar view to us in an interview. Interview with Capt. Daniel Gerard, Cincinnati Police Dep't, in Cincinnati, Ohio (Jan. 23, 2014). On Operation Vortex, see DAVID M. KENNEDY, DON'T SHOOT: ONE MAN, A STREET FELLOWSHIP, AND THE END OF VIOLENCE IN INNER-CITY AMERICA 232-66 (2011). 95. Interview with Lt. Mars Herald, Cincinnati Police Dep't, in Cincinnati, Ohio (Jan. 22, 2014). 96. MICHAEL S. SCOTT & KELLY DEDEL, U.S. DEP'T OF JUST., OFFICE OF CMTY. ORIENTED POLICING SERVS., STREET PROSTITUTION (2d ed. 2006). 97. POP discussions sometimes add a third category-recurring victims. Domestic violence cases would be the largest constituent. However, such cases tend to overlap with the other two categories. 98. See John Eck, What Do Those Dots Mean?: Mapping Theories With Data, in CRIME MAPPING AND CRIME PREVENTION 379-406 (David Weisburd & Tom McEwen eds., 1997) (arguing that effective mapping depends on hypotheses about the social processes producing the data). Spatial mapping Yale Journal on Regulation Vol. 33, 2016 problems identified in this manner is Crime Prevention Through Environmental Design. Interventions of this kind could involve enhancing visibility by altering landscaping or improving lighting. Other interventions aim to make the locale less attractive or convenient for undesirable activities. One part of a plan used to disrupt an open-air drug market in the Kennedy Heights section of Cincinnati was the attachment to a bridge siding of egg-shaped structures that made it uncomfortable to sit there. Fencing below the bridge made that space unavailable for hiding drugs.99 Markets for drugs or prostitution can sometimes be disrupted by rerouting traffic, for example, by making the route from the highway to a local street corner more circuitous. The police counsel shop owners on ways to display their merchandise that inhibit shoplifting. Some POP strategies coordinate crime control with regulation of real property use. Where rental properties are used as bases for drug dealing or prostitution, the police may pressure the landlord to evict the wrongdoers. Where neglected or abandoned property is attracting criminal activity, the police may induce building code or public nuisance enforcement to force the owner to improve conditions, or, in extreme cases, to forfeit the building.100 The department has an education program to help landlords screen prospective tenants for illegal activity and to identify and respond to such activity when it occurs. Empty buildings have unhealthy neighborhood effects, including attracting crime. The department therefore tries to work with developers to stabilize or renovate buildings critical to its public safety strategies. District 3 recently developed a strategy of this kind for the East and Lower Price Hill neighborhoods in collaboration with neighborhood activists, the city building and health departments, and the Port of Cincinnati.' 0 1 Public officials also use liquor law enforcement strategically. Bars associated with public drunkenness are likely to be threatened with loss of their liquor licenses if they do not become more careful about refusing to serve intoxicated patrons. One initiative mitigated a problem of chronic disorderly behavior at a particular street comer by persuading a local convenience store to stop selling beer in forty-ounce containers.

in AP is typically based on jurisdictional boundaries or abstract statistical properties. See JOHN ECK ET AL., U.S. DEP'T OF JUST., NAT'L INST. OF JUST., MAPPING CRIME: UNDERSTANDING HOT SPOTS (2005). 99. 2006 PROBLEM SOLVING REPORT, supranote 64, at 22. 100. Strategies of this kind are used in many cities, including New York. Building and liquor law enforcement are the principal non-confrontational strategies commonly mentioned in the Compstat literature. 101. For a more extensive account of a highly-regarded POP initiative that combined policing and economic development, see the report by Kansas City's police chief on the city's response to disorder at day labor hiring sites. The city and local stakeholder groups organized and built a center that provides services to workers and employers while regulating the hiring process in ways that limit traffic disruption and criminal activity. See James Corwin, Day Laborers: Improving the Quality of Life for Laborers, Employers, and Neighbors, THE POLICE CHIEF, Apr. 2006. Police Accountability

The most prominent responses to problems defined in terms of offenders, as opposed to places, are developed through CIRV.1 °2 CIRV begins with an effort to identify violent offenders and their social networks. Its strategies emphasize four basic elements: (1) credible threats directed to recurrent law-breakers of harsh sanctions in the event of further violence, (2) efforts to mobilize peer pressures by threatening sanctions against the group if any member offends, (3) moral exhortation by community leaders, and (4) offers of social services to help with medical or psychological problems or to improve access to employment. The most distinctive practice is the "call in." Gang members are "invited" (for those on probation or parole, the invitation is mandatory) to meetings where teams of police, prosecutors, community leaders, and social service providers deliver the combination of threats, exhortations, and offers of help. To be credible, the threats must be targeted at known wrongdoers, and, ideally, accompanied by demonstrations that the authorities have enough evidence to prosecute them for past misdeeds should they reoffend. (The teams sometimes show videotapes at the call-ins of audience members engaged in drug dealing or vandalism). One reason why targeted threats are more credible then generalized ones is that offenders know that authorities do not have the resources to follow through on the latter. But they can follow through on targeted threats, and of course, to maintain credibility, they must do so. Thus, arrest and prosecution, sometimes for minor offenses, is a key part of CIRV, but they are used only as a last resort and only against the persistently violent. The basic POP model prescribes that, after the initial intervention is deployed, its efficacy must be assessed and, if necessary, the intervention recalibrated. In principle, a good initial plan should specify measures of efficacy, though these may have to be revised as understanding of the problem changes. Crime reports and calls for service are usually key measures, but others, including customized ones, are feasible. The most systematic assessments have been performed in connection with CIRV. They have shown at least modest success, measured by gang-related violent incidents with appropriate controls. But they have also raised questions about the effectiveness of particular practices. Social services have been a particular concern. When indicators suggested services were not contributing to the reductions in violence, the program adjusted. It intensified service monitoring through a multi-agency committee that meets monthly. And it reconfigured the menu of services, focusing less on job readiness and more on psychological issues such as anger management or interpersonal coping. It also sought to target 10 3 the offers more precisely at individuals with strong propensities for violence.

102. See Engel et al., supra note 64. 103. Engel et al., supra note 64; UNIV. OF CINCINNATI POLICING INITIATIVE, IMPLEMENTATION OF THE CINCINNATI INITIATIVE TO REDUCE VIOLENCE (CIRV): YEAR 2 REPORT 4-6 (2009) [hereinafter SECOND YEAR REPORT]; see also ANTHONY A. BRAGA & DAVID L. WEISBURD, CAMPBELL COLLABORATION, THE EFFECTS OF "PULLING LEVERS" FOCUSED DETERRENCE STRATEGIES Yale Journal on Regulation Vol. 33, 2016

The problem-oriented approach pushes Cincinnati toward a more decentralized administration than New York-style AP. The design of contextual strategies depends on information from beat officers and local residents. These people, David M. Kennedy emphasizes:

[K]now who is in what street groups, and who is fighting with whom; what last year's antecedent to yesterday's shooting was; who is committing the drug robberies that are not even being reported to the police; who is selling drugs on the comers; what mid-level dealer is running the crack house operated only by juveniles; what turf is claimed by which groups and who is allowed there and who is not; which domestic violence offenders are currently dangerous to what 04 women. 1

Kennedy notes that other policing regimes "generally make little use of this frontline knowledge, partly because it is often of no use in making cases and partly because ... top-down management" inhibits access to it.t 05 The precise allocation of responsibility among different levels of administration is still a matter of discussion. All but the most local problems require initiative on the part of division commanders or central staff. Yet a premise of the Collaborative Agreement was that officers at all levels should be capable of participating in problem-solving efforts. Under the Agreement, all new recruits were trained in problem-solving methods. In principle, problem- solving ability is one of the criteria on which officers are evaluated. Because of fiscal constraints, there were no new recruits from 2008 through 2014, and as we have noted, the culture of problem-solving languished for some years. Nevertheless, at least some patrol officers identify with POP. This is especially true of the "neighborhood officers" with specific local beats. One such officer recently initiated a project to disrupt a local drug market by re-designating some streets as one-way. As part of an initiative to address cellphone thefts, he drafted and helped secure passage of an ordinance requiring pawnshops that accept 106 cellphones as collateral to report them to the police. The Collaborative Agreement embodied an ambitious conception of community engagement that has not been fully realized but nevertheless continues to influence practice. The agreement was formulated in the course of a series of open meetings and roundtables orchestrated by an expert in the mediation of civic disputes, and this type of engagement continues. In recent years, the department has conducted periodic "town hall" meetings in each of the t0 7 city's five divisions.

ON CRIME (2012) (meta-analysis concluding that nine of ten relevant studies show significant effects for "focused deterrence" interventions, though also noting that the studies lack rigor). 104. Kennedy, supra note 65, at 160. 105. Id. 106. Interview with Officer David Epstein, Cincinnati Police Dep't, in Cincinnati, Ohio (Jan. 24, 2014). 107. On the community engagement that preceded the settlement, see Eck & Rothman, supra note 90; on town hall meetings, see Cincinnati Police Department to Hold District Town Hall Police Accountability

In addition to this relatively passive participation, local stakeholders often play roles in formulating and implementing specific strategies. The department has worked with social service agencies to develop assistance for prostitutes open to exploring other means of supporting themselves. Job-related services are an important part of the CIRV violence reduction strategy. Some strategies incorporate efforts to increase the presence of law-abiding citizens at strategic times and places. Community groups may agree to turn out members as part of strategies to evict disorderly activity from a contested public space. The department has a Citizens on Patrol program in which volunteers observe 108 designated locations and report problems. A remark in one of our interviews suggests how the reconception of policing promoted by POP parallels and converges with changes in other fields. In discussing his work with a community development organization in Cincinnati's Walnut Hills neighborhood, Captain Daniel Gerard noted that he saw similarities between this work and that of a friend involved in economic and institutional development efforts as an army officer in Helmand Province, Afghanistan. The implication of the comparison is that POP more closely resembles the counter-insurgency model of warfare associated with General David Patraeus than General Patton's mobile tank tactics invoked by Bratton to explain Compstat. Like POP, the counter-insurgency approach prescribes that patrol, response to incidents, and use-of-force be coordinated with diverse, proactive initiatives that engage civilians with a stake in achieving security. The goal is to secure terrain by building a viable community, not by attempting to annihilate all potentially hostile forces. POP-influenced police offers often say "we couldn't arrest our way out of this problem." Likewise, David Petraeus often said in Iraq that "we would not be able to kill or capture our way out of" problems 09 there.'

b. Limitations

Since POP has rarely been rigorously and comprehensively implemented, it is difficult to separate limitations that arise from inadequate implementation

Meetings, CtNCINNATI-OH.GOV, http://www.cincinnati-oh.gov/police/news/cincinnati-police- department-to-hold-district-town-hall-meetings (last visited Oct. 31, 2015). 108. Officers have cultivated ties with organized community interlocutors. The District 3 commander sat on the board of the Walnut Hills Redevelopment Foundation. Another district commander helped organize and sat on the board of the Faith Community Alliance of Greater Cincinnati, a group of clergy, community leaders, and social service providers that "serves as a conduit between the community and local government." Interview with Capt. Daniel Gerard, supra note 94; interview with Capt. Gary Lee, Cincinnati Police Dep't, in Cincinnati, Ohio (Jan. 24, 2014). 109. David Petraeus, Forewordto PETER R. MANSOOR, SURGE: MY JOURNEY WITH DAVID PETRAEUS AND THE REMAKING OF THE IRAQ WAR ix, xxii (2013). Two other Cincinnati officers- Captain Maris Herald, a former social worker, and Sergeant Julian Johnson, a former school teacher- compared their practice to social work. This is the same analogy Mayor Giuliani used, but for these officers it had positive rather than disparaging connotations. The positive connotations were associated with the problem-solving orientation that has long been a central tradition in social work. Yale Journal on Regulation Vol. 33, 2016 from those that are inherent in the model. Nevertheless, looking at the theoretical accounts, we find two broad limitations. They involve ways in which the model is incomplete, though perhaps not irremediably so. First, POP's commitment to multiple and flexible criteria of success is both a strength and a weakness. It facilitates more complex responses. However, it also makes it harder to evaluate success. Since criteria will vary across different initiatives, rigorous comparisons will be difficult. And since criteria are provisional, it may be unclear whether a low score reflects the inadequacy of the intervention or the inadequacy of the metric. This problem can be mitigated by incorporating some basic standard measures such as crime rates, but moving in this direction compromises the ambition to contextualize. A common response in other fields is to adopt modes of evaluation that assess process as well as outcomes and employ qualitative as well as quantitative judgments. For example, the "balanced scorecard" used in many fields summarizes both quantitative measures and qualitative judgments on both process and outcomes. Process variables in POP would include the quality of problem definition, planning, and stakeholder engagement. The qualitative judgments typically emerge from a peer review process in which outsiders with relevant experience audit samples of cases." 0 Such processes, however, are severely underdeveloped in policing. Second, there is no clear model of how local problem-solving efforts fit within the larger structure of policing. Larger structures above the local problem level have to deal with three sorts of issues. First, they need to collect and analyze information about local experiences in forms that permit learning. Second, they need to prioritize problems and allocate resources within the jurisdiction. Third, they need to coordinate activities across neighborhoods, and indeed across cities, states, and nations for problems that reach across jurisdictions. POP proponents have made most progress on the first, learning-facilitation goal. I I In Cincinnati the CIRV reorganization made substantial progress on these issues for the violence reduction programs. The reorganization involved sophisticated assessment instruments for both process and outcomes, and it produced an explicit and comprehensive organizational structure with clearly assigned responsibility for data gathering and reassessment.11 2 On the other hand, problem-solving efforts outside of CIRV are less systematically coordinated and assessed in Cincinnati.

110. See ROBERT KAPLAN & DAVID NORTON, THE BALANCED SCORECARD: TRANSLATING STRATEGY INTO ACTION (1996). We have described such a review process used in child protective services in Kathleen G. Noonan et al., LegalAccountability in the Service-Based Welfare State: Lessons from Child Welfare Reform, 34 LAW & SOC. INQUIRY 523, 523-68 (2009). Ill. A preliminary effort to address both of our problems appears in RACHEL BOBA & ROBERTO SANTOS, U.S. DEP'T OF JUST., OFFICE OF CMTY. ORIENTED POLICING SERVS., A POLICE ORGANIZATIONAL MODEL FOR CRIME REDUCTION: INSTITUTIONALIZING PROBLEM SOLVING, ANALYSIS, AND ACCOUNTABILITY (2011). 112. Tillyer et al., supranote 90, at 979-82, 986-89; SECOND YEAR REPORT, supranote 103. Police Accountability

IV. Civil Rights and Police Reform

Post-bureaucratic organization is distinctively responsive to the difficulties of second-generation civil rights doctrine. Those difficulties, we have seen, include applying the notion of intent to disparate harm that results from inattention and applying the notion of reasonableness to conduct that is normatively ambiguous. Post-bureaucratic organization responds to these difficulties by insisting on and facilitating self-consciousness and articulation. Indeed, much of the literature on post-bureaucratic organization portrays these qualities, not just as instruments to greater productivity, but as intrinsic values that constitute a kind of organizational virtue." 3 As applied to contemporary civil rights cases, they imply a duty to examine rigorously the effects of conduct on civil rights values and to resolve ambiguity by articulating provisionally but reflectively the organization's understanding of issues that have not been resolved externally. The judicially induced reforms in New York and Cincinnati illustrate the emerging duty of responsible administration. Their concrete directives owe more to norms of administrative practice than to any interpretive inferences from substantive equal protection or search and seizure norms. Yet, the two regimes reflect a basic difference in remedial design. The remedial order in New York leaves the city's core policing practices alone. It adds a set of peripheral compliance routines designed to minimize the threat of these core practices to civil rights values. The Cincinnati Collaborative Agreement requires a more comprehensive restructuring. While this holistic approach is unusual in police litigation, it is part of a larger class of holistic civil rights reforms in diverse fields. Comprehensive reform sometimes appears more efficient than specialized compliance procedures from the perspectives of both civil rights and core crime-control goals.

A. Compliance

Because the city did not settle until the case was on appeal, the New York lawsuit produced a rare contested judgment. The trial court was thus compelled to address substantive doctrinal concerns, but its remedial order is generally consistent with the PTSR approach. As such, it illustrates both the strengths and weaknesses of that approach. Whether the plaintiffs established systemic violations under current doctrine is debatable, and the court's decision would have been vulnerable had

113. On self-consciousness, see FRED KOFMAN, CONSCIOUS BUSINESS: HOW TO BUILD VALUE THROUGH VALUES (2012); and on articulation, see PRODUCTIVITY PRESS DEVELOPMENT TEAM, STANDARD WORK FOR THE SHOPFLOOR 1-12 (2002). Yale Journal on Regulation Vol. 33, 2016 the appeal gone ahead.1 14 We think the strongest case for the court's decision can be made in terms of a duty of responsible administration. The evidence showed extensive administrative neglect or incompetence with respect to policy-making, supervision, and monitoring of civil rights norms: the city knew that the document it used for many years to record stops for civil rights monitoring purposes was inadequate but had made no effort to revise it. It was aware of long-term and widespread non-compliance with frontline documentation requirements but failed to address it. It apparently made no use for monitoring purposes of the records it had been required to keep under an earlier decree. Its only detailed written policies on racial profiling and the constitutional limits on search and seizure were in training manuals, and the court found them erroneous. Although the policies prohibited racial profiling in general terms, there appeared to be little or no supervisory effort to elaborate or enforce them. The department used the quantity of stops and arrests that an officer made as a factor in evaluations, and many officers felt pressure to get their numbers up without regard to whether the stops were lawful. Although the department recognized that quotas were inappropriate (and, beginning in 2010, prohibited by city ordinance), it never produced a coherent written or oral statement explaining how the quantity of stops figured into evaluation." 5 Within the department, discipline for civil rights violations was rare in part because the responsible officials, in violation of the department's regulations, rejected all complaints that depended on uncorroborated civilian testimony disputed by the officer named in the complaint. Despite numerous claims of systemic racial bias over more than a decade, including a 1999 New York Attorney General report, virtually the only rigorous effort to monitor equality norms was a 2007 RAND Report. That report concluded, on a basis the court found questionable, that aggregate data did not suggest discrimination. However, it also concluded that outlier data in particular localities raised questions that

114. The plaintiffs' case was extensive, but it rested mainly on four points: (1) On discrimination, the plaintiffs introduced evidence of statements by senior officers as direct evidence of purposeful discrimination; (2) as indirect evidence, they introduced a sophisticated statistical analysis by our colleague Jeffrey Fagan showing large disparities between the racial composition of the people stopped, frisked, and arrested and that of the residents of the neighborhoods in which the police activity occurred; (3) on search-and-seizure, they introduced testimony about nineteen stops of which the court found nine to be unlawful; and (4) they also offered a study by Fagan concluding on the basis of an examination of the department's records that at least 200,000 of 4.2 million stops between 2004 and 2012 were unlawful. Floyd v. City of New York, 959 F. Supp. 2d 540, 572-58 (S.D.N.Y. 2013). An appellate court hewing to the precepts of classicism might have concluded: (1) most of the official statements are too ambiguous to show purposeful discrimination; (2) the statistical study is strong, but it depends on complex and disputed methodology of a sort that appellate authority resists in the criminal justice context; (3) the nine individual instances of Fourth Amendment violations look like the showing that Rizzo held inadequate; and (4) the search and seizure study is based on sloppily kept, unreliable records. (Of course, if there is a duty of responsible administration, this latter point is unhelpful to the city. But classical doctrine sometimes rewards administrative laxness. See supra note 24 and accompanying text). 115. Floyd, 959 F. Supp. 2d at 589-624. Police Accountability

should be investigated, and the department failed to follow up on this or other 16 recommendations in the report.' In important respects, the court's remedial order seems more directly responsive to these administrative failings than to its findings on discriminatory intent or objective unreasonableness. For the most part, it follows the PTSR approach of the consent decrees. 1 7 The central remedial intervention is the appointment of a monitor "to develop, based on consultation with the parties, a set of reforms of the NYPD's policies, training, supervision, monitoring, and discipline regarding stop and frisk." '118 The court laid down some specific parameters for documentation of stops, and it ordered the department to "experiment" with using body-worn cameras to videotape stops in at least one precinct in each of the five boroughs. In addition, the court appointed a second judicial officer called a "facilitator" to guide a "joint remedial process." '19 The court's order avoids the danger of judicial micro-management, but it does reflect limitations common to PTSR-style remedial practice. These limitations reflect the lingering influence of classical doctrinal notions. With respect to discrimination, we have noted that, under an ambitious conception of non-discrimination, the defendant has a duty to investigate and assess the disproportionate costs its practices impose on protected groups and to consider ways in which these harms might be mitigated. However, in the court's analysis, both the harms and the practices in question are defined narrowly. The court is most concerned with the harms to law-abiding minority group members. It does not treat as justiciable, or at least as remediable, claims arising from the mass criminalization of young minority men through aggressive minor- offense enforcement, even though this harm is racially skewed and viewed by

116. Id. at 590-91, 622-24. Another pertinent indication of managerial irresponsibility not mentioned in the court's opinion is the city's treatment of information from lawsuits. Hundreds of lawsuits are filed yearly against the police-there were 2,211 filed in fiscal year 2006-but up to the time of the stop-and-frisk challenge the department made no managerial use of the claims or information developed in the suits. It did not examine or investigate the facts unless the plaintiffs choose to file separate administrative complaints. It did not record judicial claims in the personnel files of the officers named in them. The law department provided almost no information to the police about the claims. These practices persisted for many years despite recommendations by three city comptrollers that the police department examine lawsuit claims diagnostically. Joanna C. Schwartz, Myths and Mechanics of Deterrence. The Role of Lawsuits in Law Enforcement Decisionmaking, 57 UCLA L. REV. 1023, 1045-48 (2010). The city recently announced an initiative along these lines. Benjamin Weiser, ComptrollerAims to Curb Personal- Injury Claims Against New York City, N.Y. TiMES, July 9, 2014, http://www.nytimes.com/2014/07/09/nyregion/comptroller-aims-to-curb-personal-injury-claims-against- new-york-city.html. 117. Floyd, 959 F. Supp. 2d at 668. Some commenters worry that the process of negotiated settlement subverts public control and inappropriately preempts the law-declaring function of the courts. See, e.g., Samuel Issacharoff, When Substance Mandates Procedure: Martin v. Wilks and the Rights of Vested Incumbents in Civil Rights Consent Decrees, 77 CORNELL L. REV. 189, 238-39 (1992). But as New York exemplifies, trial court remedial practices in contested cases are likely to parallel practice in settlements. When they have to prescribe remedies, judges will look to remedies in comparable situations rather than trying to derive them formally from substantive doctrine. 118. Floyd, 959 F. Supp. 2d at 678. 119. ld. at 966-88. Yale Journal on Regulation Vol. 33, 2016 many as more serious than harm to the law-abiding. The court does not recognize a duty on the part of the city to examine the efficacy of stop-and-frisk as a 20 strategy for controlling severe crime or to search for a less harmful alternative. The court's refusal to take the more ambitious course is consistent with doctrine that cautions against drawing strong inferences from the disparate impacts of wholesale choices about which laws to enforce, 21 but it is nevertheless debatable. What the court sees as a narrower intervention-a command to eliminate discrimination from the established crime-control approach-may actually prove quite difficult to enforce once the city has coached its officers to avoid inculpatory statements and facially inadequate documentation. A more encompassing intervention might have been easier to implement. It would not, as the court might have assumed, require a ban of stop- and-frisk, much less prescription of an acceptable alternative. Rather, it would have required the department to expand the self-analysis and stakeholder deliberations the court ordered to include a disciplined consideration of alternative crime-control approaches to see if there is a comparably effective but less harmful way of achieving the city's goals. With respect to search and seizure, we have noted the problem that, while Fourth Amendment constitutional duty applies most importantly to the city as an institution, doctrine and practice tends to be elaborated from the perspective of the individual police officer at the moment of intervention. A strategy using massive arrests for minor offenses as a way of gaining leverage over major offenses and tolerating low hit rates on stops implies relative ignorance about the community (other than the location of reported crimes). A department that invests effectively in cultivating useful community relations, in aggregating and analyzing intelligence, and in facilitating efficient exchange of information among its members may be able to target its interventions more precisely. New York City claims that it does these things, but its rigid commitment to stop-and- frisk and its low hit rates raise doubts about the rigor of its efforts. When it succeeds, POP makes more sparing use of arrests but achieves high hit rates because it is able to reliably identify the actors most responsible for community disorder. Many believe that, contrary to the apparent assumptions of some AP proponents, a very small number of actors are responsible for a large fraction of disorder even in the most disordered communities.122 To the extent

120. "1 emphasize at the outset ... that this case is not about the effectiveness of stop and frisk in deterring or combating crime."Id. at 556. 121. Atwater v. City of Lago Vista, 532 U.S. 318, 350-51 (2001); Whren v. United States 517 U.S. 806, 818 (1996). 122. For example, in an effort to clean up a severely crime- and drug-ridden neighborhood in Austin, Texas, intensive intelligence analysis indicated to the surprise of some that there were only sixteen active dealers. According to one official, "This exercise helped officers realize that they may have been directing enforcement action toward individuals who lived in and around the drug market but who were not actually involved in it." Kennedy, supra note 65, at 168. See also HUMAN RIGHTS WATCH, supra note 78, (study of New York City records showing that only 3.1 percent of those arrested for marijuana possession in 2003 and 2004 were subsequently convicted of a violent felony). Police Accountability that this is so, AP is a highly inefficient way of addressing it. Arguably, the availability of less harmful alternatives should bear on reasonableness under the Fourth Amendment, as it does under some antidiscrimination and tort doctrines.'23 Departments using AP should thus have a duty to examine alternatives, such as POP, and to either adopt one or provide credible reasons for why it would be less satisfactory than current practices. The court's insistence that efficacy is not at issue removes pressure from the department to question the premises of its strategy. In addition, the New York order, consistent with Whren v. United States, understands Fourth Amendment reasonableness to depend on the extent to which evidence supports the officer's judgment that some law (no matter how minor) has been broken, not on the extent to which intervention furthers underlying crime-control purposes. Critics object that Whren leaves arbitrariness unchecked. Defenders of the case warn of the dangers of a standard that turns on the subjective intentions of individual officers. But from our perspective, the key focus of inquiry should not be the subjectivity of the individual officer, but the departmental strategy that the officer's conduct implements. Can the department show that it has a strategy that explains its officers' practices? Is it assessing the efficacy of these practices in light of its own experience and those of comparable agencies? If not, then it is violating its duty of responsible administration. The limits of classic substantive doctrine explain the limits of the remedial order, but it is not clear that the authority requires the limits. Remedial practice has emancipated itself from the constraints of classic substantive doctrine in many ways. Perhaps it might go further. Cincinnati, to which we return in the next section, suggests that under some conditions, it could.

B. The HolisticApproach

1. Limits of Specialized Reform

Sometimes the most effective way to vindicate civil rights values is to change an institution's core practices. Sometimes institutions can be persuaded or induced to undertake such far-reaching changes because they turn out to be less costly than peripheral ones. There are two general reasons why comprehensive reform may prove more effective. First, compliance procedures added to unreformed core processes may prove too weak to affect practice or may generate costly friction. Power, honor, and reward in an organization tend to go to those who excel at attaining core

123. On antidiscrimination, examples include the less-restrictive-alternative and reasonable-accommodation principles, see supra notes 32 and 43; on tort, see RESTATEMENT (THIRD) OF TORTS: PRODUCTS LIABILITY § 2(b) (1998) (providing that negligence liability for defective products depends in substantial part on whether "harm posed by the product could have been reduced or avoided by the adoption of a reasonable alternative design"). Yale Journal on Regulation Vol. 33, 2016 goals. Commands that impede the attainment of core goals may be perceived as illegitimate or hypocritical, and those charged with enforcing them may be seen as scolds or snitches. Reforms that reduce the tension between core goals and civil rights norms can reduce such difficulties. For example, many American producers of consumer products commit to induce their foreign suppliers to comply with international labor standards. The track record of these efforts, despite the good faith of most monitors and substantial investments by many producers, has been poor. Monitors find it hard to get information on remote, globally-dispersed operations, and they have only weak influence over the producers' subcontracting decisions. Yet, some observers see promise in a recent initiative by Nike, the sports apparel producer, to shift its foreign suppliers' core manufacturing to "lean production."' 124 Lean production is attractive to global producers because it enables them to respond more quickly to changes in demand. Going lean requires extensive worker training, and such an investment tends to make the workers valuable assets. A rigorous analysis of Nike labor-standard audit records shows that subcontractors in the program that successfully move to lean production have dramatically better records than their traditional peers on such important matters as minimum wages, 25 abuse of overtime, and underage labor.1 A second reason why specialized intervention may fail is that poor respect for civil rights may be a symptom of broader organizational dysfunction. Lani Guinier and Gerald Torres suggest that evidence of civil rights violations is often a kind of "miner's canary" that signals more pervasive pathology.' 26 Efforts to reform a discrete piece of a generally failing organization may be thwarted by surrounding dysfunction. In such situations, it may turn out that comprehensive change can serve other institutional goals as well as civil rights. Efforts to address race discrimination in juvenile justice provide an illustration. Minority youth are overrepresented at every stage of the criminal justice process, especially incarceration. Amendments in 1992 and 2002 to the Juvenile Justice and Delinquency Prevention Act mandate that juvenile justice officials proactively address "disproportionate minority contacts."'' 2 7 They must

124. See RICHARD M. LOCKE, THE PROMISE AND LIMITS OF PRIVATE POWER: PROMOTING LABOR STANDARDS IN A GLOBAL ECONOMY (2013). 125. Greg Distelhorst et al., Does Lean Improve Labor Standards?: Capability Building and Social Performance in the Nike Supply Chain, 61 MANAGEMENT SCIENCE (forthcoming 2016); see also Charles F. Sabel, On Richard M. Locke, The Promise and Limits of Private Power: Promoting Labor Standards in a Global Economy, 12 SOCIO-ECON. REV. 219, 226-35 (2014). The improved performance does not spill over into health-and-safety compliance, presumably because workers are less attentive to such concerns until some salient harm occurs. Consider also Jennifer Hochschild's assessment of school desegregation, which argues that the more successful efforts tended to involve comprehensive reforms that reached core educational practices, such as replacing competitive with collaborative classroom practices and reducing ability grouping. JENNIFER L. HOCHSCHILD, THE NEW AMERICAN DILEMMA: LIBERAL DEMOCRACY AND SCHOOL DESEGREGATION 83-88 (1984). 126. LANI GUINIER & GERALD TORRES, THE MINER'S CANARY: ENLISTING RACE, RESISTING POWER, TRANSFORMING DEMOCRACY (2003). 127. Johnson, supra note 3, at 402; 42 U.S.C. § 5633(a)(22) (2000). Police Accountability measure and report on racial incidence, develop plans to mitigate disproportionate effects, and periodically reassess these plans. The Department of Justice and the Annie E. Casey Foundation have developed a national structure that provides technical assistance and facilitates peer exchanges among state and 28 local governments. The dominant strategy that has emerged among the participants in the Casey Foundation network is a reform of the entire detention process. A key element involves the replacement of informal probation-officer judgments at the entry stage with a risk assessment instrument that dictates decisions on the basis of numerical scores for key variables. The instruments are constructed through statistical analysis of past experience to determine which indicators predict reoffense or failure to appear at a hearing. Scores can be overridden, but only with supervisory approval, and the overrides are periodically reviewed to see if they indicate a need for changes in the instrument. The reforms have been associated with large reductions in the percentage of youth of all races detained.2 9 Although aggregate racial disparities have not changed much, there have been notable local reductions, and youth of color have benefitted greatly from the aggregate detention reductions. So far, the story seems a striking example of the "miner's canary" idea that racial disparities can signal a more general problem. For our purposes, the key point is that the reformers plausibly concluded that the most effective way to vindicate civil rights was to reform the institution's core processes, rather than to add a 130 specialized compliance function.

2. Problem-Oriented Policing as a Civil Rights Remedy.

Many police regimes that have been subject to structural challenges involve one or both of the conditions that favor holistic intervention. Their crime control

128. Johnson, supra note 3; Sabel & Simon, supra note 3, at 21-27. 129. RICHARD A. MENDEL, JUVENILE DETENTION ALTERNATIVES INITIATIVE, TWO DECADES OF JDAI: A PROGRESS REPORT 14-24 (2009). 130. Consider also the transformation of the idea that a child has a right to an education that takes reasonable account of her individual needs from a specialized category to a universal one. Federal statutes in 1975 and 1990 mandated that children with learning disabilities be given tailored services through an "Individualized Education Program" formulated by professionals in consultation with their parents, The program has benefited millions of children, but its eligibility limits have come to be seen as vague and arbitrary. See MARK KELMAN & GILLIAN LESTER, JUMPING THE QUEUE: AN INQUIRY INTO THE LEGAL TREATMENT OF STUDENTS WITH LEARNING DISABILITIES (1997). Reforms from the No Child Left Behind Act through various initiatives of the Obama administration have sought to induce individualized attention to all students, and especially persistently struggling ones, regardless of whether they have a medical diagnosis. See 20 U.S.C. § 6301 (2006) (declaring the right of "all children" to reach minimum proficiency and mandating remedial services for students in persistently lagging schools); Martin Kurzweil, Disciplined Devolution and the New Education Federalism, 103 CAL. L. REV. 565 (2015) (describing Obama administration initiatives). Yale Journal on Regulation Vol. 33, 2016 strategies depend on street confrontations based on overbroad and racially inflected criteria. These strategies are constitutionally problematic even where they are effective, 3 ' and officers tend to believe they are effective more than they are. Thus, specialized civil rights remediation is likely to generate friction by inhibiting what officers consider to be their core functions. Moreover, the insufficient attention to civil rights values in these regimes is often symptomatic of more general administrative underdevelopment and dysfunction. Cincinnati illustrates the possibility of litigation-induced holistic reform. The city entered into two agreements to reform its police department in 2002. One was with the Department of Justice, which initiated an investigation at the invitation of the mayor in the aftermath of civil unrest over a fatal police shooting of an unarmed man. The other was a "Collaborative Agreement" settling a lawsuit alleging racially biased policing brought by the American Civil Liberties Union, a local advocacy group called the Black United Front, and some individuals. 132 The Department of Justice agreement and part of the Collaborative Agreement focused on familiar PTSR measures. Yet the most extensive provisions of the Collaborative Agreement prescribed uniquely ambitious reform. "The City," the decree proclaimed, "shall adopt problem-oriented 33 policing as the principal strategy for addressing crime and disorder problems."1 A prefatory "Value Statement" asserts that POP "frames the overall philosophy and practices" of the Agreement. More specifically, the city agrees that "[i]nitiatives to resolve crime and disorder shall be preceded by careful problem, definition, analysis, and an examination of a broad range of solutions." Further, the police must "routinely evaluate implemented solutions." The Agreement adds, "A law enforcement response is always a possibility, but may not be required." The Agreement recites that every Cincinnati officer will receive at least eight hours of training in the SARA (Scanning, Analysis, Response, 34 Assessment) methodology.1 It further sets out administrative procedures designed to create a "'continuous learning' process." The city will develop a "system... [to] enable the tracking of repeat offenders, repeat victims, and/or repeat locations that are necessary to community problem oriented policing." The system must enable the "tracking" of problems so that progress can be ascertained and so that officers

131. See Jeffrey Bellin, The Inverse Relation Between the Constitutionality and Effectiveness of New York City "Stop and Frisk, " 94 B.U. L. REv. 1495 (2014) (arguing that stop-and- frisk "might" work as a crime-control measure by reducing gun carrying, but only because of the unconstitutional overbreadth of its criteria). 132. See Collaborative Agreement, supra note 92; U.S. DEP'T OF JUST., MEMORANDUM OF AGREEMENT BETWEEN THE UNITED STATES DEPARTMENT OF JUSTICE AND THE CITY OF CINCINNATI, OHIO AND THE CINCINNATI POLICE DEPARTMENT (2002); Eck & Rothman, supra note 90. 133. Collaborative Agreement, supra note 92, at 16. 134. Id. at 2, 20-24. Police Accountability can identify interventions that may have useful lessons for related or similar problems. The city committed to examine practices in other jurisdictions and 135 maintain a "library of best practices." The city also committed to develop "ongoing community dialog and interaction." In addition to the complaint and monitoring processes, the agreement provided for "periodic surveys" of samples of both citizens and police officers on their views of police-citizen relations.'36 The Agreement implicitly rejects the premise of substantive Fourth Amendment doctrine that probable cause for a stop or arrest establishes its reasonableness. The SARA process assesses reasonableness in terms of the broader problem-solving plan that the action implements. At the same time, the Agreement is emphatic in rejecting intent as a touchstone. A "[c]entral" premise is that "blame is an obstacle to progress." Blaming, the document asserts, distracts attention from the common interests different groups share and from the search for mutually beneficial practices.1 37 The implication is that the racially disproportionate harm from the city's policing practices results, not from intent, but from indifference or ignorance, and that the appropriate remedy is to realign attention, not incentives, and to induce learning. The Collaborative Agreement has run its term (initially five years, extended to six), but as we have seen, it continues to influence practice in Cincinnati, albeit incompletely. Some citizens and officers continue to refer to "the Collaborative" as an ongoing institution. The embrace of POP as a civil rights remedy is partly due to some historical accidents.' 38 However, the appeal of POP as a civil rights remedy rests on broader considerations. POP potentially facilitates better-informed and more nuanced decisions that affect civil rights values. In particular, it has the potential to reduce the need for imprecisely targeted coercive interventions that threaten both Fourth Amendment and Equal Protection values. With respect to discrimination, POP encourages a more continuous and localized assessment of disproportionate harm to protected groups and a search for less harmful alternatives. With respect to search and seizure, it potentially facilitates more focused strategies for controlling serious crimes that rely less on vague and overbroad criteria for intervention. POP respects the injunction of Terry v.

135. Id. at 29c, 29p, 29b. Observation on our visit in January 2014 suggested that implementation of these provisions had fallen off. 136. Id. at 29f, 34. 137. ld.at 2. 138. The negotiations took place in an atmosphere of crisis in which the city had been roiled by shocking shootings of both an unarmed civilian and three police officers. To the extent that the community was divided into factions, each side had compelling reason to acknowledge the concerns of the other and to seek solutions that involved mutual engagement. See Eck & Rothman, supra note 90. In addition, the University of Cincinnati happened to be the professional home of John Eck, a national leader in the development of POP. Alphonse Gerhardstein, the experienced plaintiffs' lawyer well-attuned tothe limitations of traditional judicial remedies, was impressed by Eck's ideas, and the two presented them in the course of the police-community engagement that led to the Agreement. Yale Journal on Regulation Vol. 33, 2016

Ohio1 39 that police confrontation be based on reasonable, "individualized" suspicion. It intervenes more selectively and with a richer information base than AP. In addition, it strives for the kind of accountability that the Whren court considered infeasible. It asks offices to justify their decisions, not just in terms of whether some law has been broken in a particular situation, but in terms of a broader crime-control strategy. At the same time, POP extends the transparency and reasonable- explanation themes of the duty of responsible administration. It provides explanations of police conduct that are accessible to the local community. As it enables the police to profit from stakeholder knowledge, it enables stakeholders to better assess conduct both for general efficacy and for compliance with civil rights norms. It encourages citizen involvement, not just as sources of information, but as active participants who have something to contribute to the efficacy of the strategy. We do not suggest that Cincinnati's approach is constitutionally mandatory. Even if relevant doctrine is understood to entail disciplined analysis of harms and alternatives, there may be many ways of institutionalizing these practices effectively. It is possible that a more centralized and standardized regime might do so. But POP seems more directly responsive to the challenges of the emerging duty of responsible administration than any version of AP established to date. 40

V. Conclusion

Second-generation civil rights problems resist substantive regulation. Thus, doctrine has focused increasingly on the structures and practices that give content to official discretion. At its most effective, intervention has encouraged post- bureaucratic trends that minimize the unreflectiveness and ambiguity that give rise to second-generation problems. Yet, as developments in policing show with particular clarity, post- bureaucratic organization takes markedly different forms with correspondingly different implications for accountability. At one extreme, exemplified by AP in New York, the organization focuses on identifying high-crime locales and rapidly mobilizing a limited set of conventional interventions within them. It decentralizes only to the extent necessary to speed redeployment of forces. At the other extreme, exemplified by POP as practiced in Cincinnati, the organization aims to make its interventions at once more effective and more

139. 392 U.S. 1 (1968). 140. POP is also potentially democracy-reinforcing in a more ambitious sense than narrower PTSR-type reform. PTSR reforms reinforce democracy by inducing substantial transparency, increasing central accountability to elected officials, and sometimes by creating special commissions to investigate complaints and assess problems. See DAVID ALAN SKLANSKY, DEMOCRACY AND THE POLICE 74-113 (2008); Rachel Harmon, Why Do We Still Lack Data on Policing?, 92 MARQUE1IE L. REV. 1119 (2013). POP aspires to go farther by facilitating more direct participation by the citizens most affected by particular problems. See Joshua Cohen & Charles F. Sabel, Directly-DeliberativePolyarchy, 3 EUR. L.J. 313, 342 (1997) (elaborating a conception of democracy emphasizing stakeholder problem-solving). Police Accountability precisely targeted by tailoring interventions to particular contexts-specifically, physical environments or social networks. To acquire the necessary information, initiative is decentralized to the lower ranks, and the department collaborates with the community at many levels. These differences in strategy matter for accountability. The strategies shape the nature and determine the frequency of the situations under which police may put constitutional values most sharply at risk. Strategies such as POP that strive to distinguish wrongdoers from others in the neighborhood, to concentrate on precisely identified criminogenic locations, and to enlist community support are less likely to produce constitutionally risky confrontations than are strategies like AP. The Constitution does not mandate POP, and it would be fruitless to impose it on a department that is hostile to it. It is possible that a more centralized and standardized regime might achieve equivalent or better performance. The role of the court or the supervisory agency like the DOJ under a duty of responsible administration is not to prescribe solutions. Rather, it is to induce entities that have violated constitutional norms to undertake disciplined self-analysis of the extent and underlying causes of the harms they have inflicted and a painstaking search for less burdensome alternatives. Although the Cincinnati reforms are unusual in their comprehensiveness, core elements of ongoing, transparent self-assessment are entering best-practice conventions. The DOJ's "Principles for Promoting Police Integrity"--a starting point for remedial design in many interventions-demand an open-ended inquiry into the underlying causes of impermissible behavior by prescribing that the "precipitating events" that lead to use-of-force, searches and seizures, and other such actions should be reviewed to determine "whether any revisions to training or practices are necessary."' 41 The Early Intervention systems, which operate in accordance with the SARA principles at the core of POP, have enlarged their focus from (groups of) at risk officers to breakdowns in supervision that tolerate or encourage misconduct. From there it is a manageable step to consideration of the strategies that shape the tasks and incentives of supervisors. Cincinnati gives an imperfect but suggestive illustration of what it would mean to apply such assessment, not just in reaction to instances of malfeasance, but proactively to the agency's core crime-control strategies.

141. U.S. DEP'T OF JUST., PRINCIPLES FOR PROMOTING POLICE INTEGRITY, supra note 57, at 23.